diff options
Diffstat (limited to 'src/mesa')
170 files changed, 5673 insertions, 2294 deletions
diff --git a/src/mesa/drivers/common/meta.c b/src/mesa/drivers/common/meta.c new file mode 100644 index 00000000000..a8db686573f --- /dev/null +++ b/src/mesa/drivers/common/meta.c @@ -0,0 +1,1248 @@ +/* + * Mesa 3-D graphics library + * Version: 7.6 + * + * Copyright (C) 2009 VMware, Inc. All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +/** + * Meta operations. Some GL operations can be expressed in terms of + * other GL operations. For example, glBlitFramebuffer() can be done + * with texture mapping and glClear() can be done with polygon rendering. + * + * \author Brian Paul + */ + + +#include "main/glheader.h" +#include "main/mtypes.h" +#include "main/imports.h" +#include "main/arrayobj.h" +#include "main/blend.h" +#include "main/bufferobj.h" +#include "main/depth.h" +#include "main/enable.h" +#include "main/macros.h" +#include "main/matrix.h" +#include "main/polygon.h" +#include "main/scissor.h" +#include "main/shaders.h" +#include "main/stencil.h" +#include "main/texobj.h" +#include "main/texenv.h" +#include "main/teximage.h" +#include "main/texparam.h" +#include "main/texstate.h" +#include "main/varray.h" +#include "main/viewport.h" +#include "shader/program.h" +#include "swrast/swrast.h" +#include "drivers/common/meta.h" + + +/** + * State which we may save/restore across meta ops. + * XXX this may be incomplete... + */ +struct save_state +{ + GLbitfield SavedState; /**< bitmask of META_* flags */ + + /** META_ALPHA_TEST */ + GLboolean AlphaEnabled; + + /** META_BLEND */ + GLboolean BlendEnabled; + GLboolean ColorLogicOpEnabled; + + /** META_COLOR_MASK */ + GLubyte ColorMask[4]; + + /** META_DEPTH_TEST */ + struct gl_depthbuffer_attrib Depth; + + /** META_FOG */ + GLboolean Fog; + + /** META_PIXELSTORE */ + /* XXX / TO-DO */ + + /** META_RASTERIZATION */ + GLenum FrontPolygonMode, BackPolygonMode; + GLboolean PolygonOffset; + GLboolean PolygonSmooth; + GLboolean PolygonStipple; + GLboolean PolygonCull; + + /** META_SCISSOR */ + struct gl_scissor_attrib Scissor; + + /** META_SHADER */ + GLboolean VertexProgramEnabled; + struct gl_vertex_program *VertexProgram; + GLboolean FragmentProgramEnabled; + struct gl_fragment_program *FragmentProgram; + GLuint Shader; + + /** META_STENCIL_TEST */ + struct gl_stencil_attrib Stencil; + + /** META_TRANSFORM */ + GLenum MatrixMode; + GLfloat ModelviewMatrix[16]; + GLfloat ProjectionMatrix[16]; + GLfloat TextureMatrix[16]; + GLbitfield ClipPlanesEnabled; + + /** META_TEXTURE */ + GLuint ActiveUnit; + GLuint ClientActiveUnit; + /** for unit[0] only */ + struct gl_texture_object *CurrentTexture[NUM_TEXTURE_TARGETS]; + /** mask of TEXTURE_2D_BIT, etc */ + GLbitfield TexEnabled[MAX_TEXTURE_UNITS]; + GLbitfield TexGenEnabled[MAX_TEXTURE_UNITS]; + GLuint EnvMode; /* unit[0] only */ + + /** META_VERTEX */ + struct gl_array_object *ArrayObj; + struct gl_buffer_object *ArrayBufferObj; + + /** META_VIEWPORT */ + GLint ViewportX, ViewportY, ViewportW, ViewportH; + GLclampd DepthNear, DepthFar; + + /** Miscellaneous (always disabled) */ + GLboolean Lighting; +}; + + +/** + * State for glBlitFramebufer() + */ +struct blit_state +{ + GLuint TexObj; + GLsizei TexWidth, TexHeight; + GLenum TexType; + GLuint ArrayObj; + GLuint VBO; + GLfloat verts[4][4]; /** four verts of X,Y,S,T */ +}; + + +/** + * State for glClear() + */ +struct clear_state +{ + GLuint ArrayObj; + GLuint VBO; + GLfloat verts[4][7]; /** four verts of X,Y,Z,R,G,B,A */ +}; + + +/** + * State for glCopyPixels() + */ +struct copypix_state +{ + GLuint TexObj; + GLsizei TexWidth, TexHeight; + GLenum TexType; + GLuint ArrayObj; + GLuint VBO; + GLfloat verts[4][5]; /** four verts of X,Y,Z,S,T */ +}; + + +/** + * State for glDrawPixels() + */ +struct drawpix_state +{ + GLuint TexObj; + GLsizei TexWidth, TexHeight; + GLenum TexIntFormat; + GLuint ArrayObj; + GLuint VBO; + GLfloat verts[4][5]; /** four verts of X,Y,Z,S,T */ +}; + + + +/** + * All per-context meta state. + */ +struct gl_meta_state +{ + struct save_state Save; /**< state saved during meta-ops */ + + struct blit_state Blit; /**< For _mesa_meta_blit_framebuffer() */ + struct clear_state Clear; /**< For _mesa_meta_clear() */ + struct copypix_state CopyPix; /**< For _mesa_meta_copy_pixels() */ + struct drawpix_state DrawPix; /**< For _mesa_meta_draw_pixels() */ + + /* other possible meta-ops: + * glDrawPixels() + * glBitmap() + * glGenerateMipmap() + */ +}; + + +/** + * Initialize meta-ops for a context. + * To be called once during context creation. + */ +void +_mesa_meta_init(GLcontext *ctx) +{ + ASSERT(!ctx->Meta); + + ctx->Meta = CALLOC_STRUCT(gl_meta_state); +} + + +/** + * Free context meta-op state. + * To be called once during context destruction. + */ +void +_mesa_meta_free(GLcontext *ctx) +{ + struct gl_meta_state *meta = ctx->Meta; + + if (meta->Blit.TexObj) { + _mesa_DeleteTextures(1, &meta->Blit.TexObj); + _mesa_DeleteBuffersARB(1, & meta->Blit.VBO); + _mesa_DeleteVertexArraysAPPLE(1, &meta->Blit.ArrayObj); + } + + if (meta->Clear.VBO) { + _mesa_DeleteBuffersARB(1, & meta->Clear.VBO); + _mesa_DeleteVertexArraysAPPLE(1, &meta->Clear.ArrayObj); + } + + if (meta->CopyPix.TexObj) { + _mesa_DeleteTextures(1, &meta->CopyPix.TexObj); + _mesa_DeleteBuffersARB(1, & meta->CopyPix.VBO); + _mesa_DeleteVertexArraysAPPLE(1, &meta->CopyPix.ArrayObj); + } + + if (meta->DrawPix.TexObj) { + _mesa_DeleteTextures(1, &meta->DrawPix.TexObj); + _mesa_DeleteBuffersARB(1, & meta->DrawPix.VBO); + _mesa_DeleteVertexArraysAPPLE(1, &meta->DrawPix.ArrayObj); + } + + _mesa_free(ctx->Meta); + ctx->Meta = NULL; +} + + +/** + * Enter meta state. This is like a light-weight version of glPushAttrib + * but it also resets most GL state back to default values. + * + * \param state bitmask of META_* flags indicating which attribute groups + * to save and reset to their defaults + */ +static void +_mesa_meta_begin(GLcontext *ctx, GLbitfield state) +{ + struct save_state *save = &ctx->Meta->Save; + + save->SavedState = state; + + if (state & META_ALPHA_TEST) { + save->AlphaEnabled = ctx->Color.AlphaEnabled; + if (ctx->Color.AlphaEnabled) + _mesa_Disable(GL_ALPHA_TEST); + } + + if (state & META_BLEND) { + save->BlendEnabled = ctx->Color.BlendEnabled; + if (ctx->Color.BlendEnabled) + _mesa_Disable(GL_BLEND); + save->ColorLogicOpEnabled = ctx->Color.ColorLogicOpEnabled; + if (ctx->Color.ColorLogicOpEnabled) + _mesa_Disable(GL_COLOR_LOGIC_OP); + } + + if (state & META_COLOR_MASK) { + COPY_4V(save->ColorMask, ctx->Color.ColorMask); + if (!ctx->Color.ColorMask[0] || + !ctx->Color.ColorMask[1] || + !ctx->Color.ColorMask[2] || + !ctx->Color.ColorMask[3]) + _mesa_ColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE); + } + + if (state & META_DEPTH_TEST) { + save->Depth = ctx->Depth; /* struct copy */ + if (ctx->Depth.Test) + _mesa_Disable(GL_DEPTH_TEST); + } + + if (state & META_FOG) { + save->Fog = ctx->Fog.Enabled; + if (ctx->Fog.Enabled) + _mesa_set_enable(ctx, GL_FOG, GL_FALSE); + } + + if (state & META_RASTERIZATION) { + save->FrontPolygonMode = ctx->Polygon.FrontMode; + save->BackPolygonMode = ctx->Polygon.BackMode; + save->PolygonOffset = ctx->Polygon.OffsetFill; + save->PolygonSmooth = ctx->Polygon.SmoothFlag; + save->PolygonStipple = ctx->Polygon.StippleFlag; + save->PolygonCull = ctx->Polygon.CullFlag; + _mesa_PolygonMode(GL_FRONT_AND_BACK, GL_FILL); + _mesa_set_enable(ctx, GL_POLYGON_OFFSET_FILL, GL_FALSE); + _mesa_set_enable(ctx, GL_POLYGON_SMOOTH, GL_FALSE); + _mesa_set_enable(ctx, GL_POLYGON_STIPPLE, GL_FALSE); + _mesa_set_enable(ctx, GL_CULL_FACE, GL_FALSE); + } + + if (state & META_SCISSOR) { + save->Scissor = ctx->Scissor; /* struct copy */ + } + + if (state & META_SHADER) { + if (ctx->Extensions.ARB_vertex_program) { + save->VertexProgramEnabled = ctx->VertexProgram.Enabled; + save->VertexProgram = ctx->VertexProgram.Current; + _mesa_set_enable(ctx, GL_VERTEX_PROGRAM_ARB, GL_FALSE); + } + + if (ctx->Extensions.ARB_fragment_program) { + save->FragmentProgramEnabled = ctx->FragmentProgram.Enabled; + save->FragmentProgram = ctx->FragmentProgram.Current; + _mesa_set_enable(ctx, GL_FRAGMENT_PROGRAM_ARB, GL_FALSE); + } + + if (ctx->Extensions.ARB_shader_objects) { + save->Shader = ctx->Shader.CurrentProgram ? + ctx->Shader.CurrentProgram->Name : 0; + _mesa_UseProgramObjectARB(0); + } + } + + if (state & META_STENCIL_TEST) { + save->Stencil = ctx->Stencil; /* struct copy */ + if (ctx->Stencil.Enabled) + _mesa_Disable(GL_STENCIL_TEST); + /* NOTE: other stencil state not reset */ + } + + if (state & META_TEXTURE) { + GLuint u, tgt; + + save->ActiveUnit = ctx->Texture.CurrentUnit; + save->ClientActiveUnit = ctx->Array.ActiveTexture; + save->EnvMode = ctx->Texture.Unit[0].EnvMode; + + /* Disable all texture units */ + for (u = 0; u < ctx->Const.MaxTextureUnits; u++) { + save->TexEnabled[u] = ctx->Texture.Unit[u].Enabled; + save->TexGenEnabled[u] = ctx->Texture.Unit[u].TexGenEnabled; + if (ctx->Texture.Unit[u].Enabled || + ctx->Texture.Unit[u].TexGenEnabled) { + _mesa_ActiveTextureARB(GL_TEXTURE0 + u); + _mesa_set_enable(ctx, GL_TEXTURE_1D, GL_FALSE); + _mesa_set_enable(ctx, GL_TEXTURE_2D, GL_FALSE); + _mesa_set_enable(ctx, GL_TEXTURE_3D, GL_FALSE); + _mesa_set_enable(ctx, GL_TEXTURE_CUBE_MAP, GL_FALSE); + _mesa_set_enable(ctx, GL_TEXTURE_RECTANGLE, GL_FALSE); + _mesa_set_enable(ctx, GL_TEXTURE_GEN_S, GL_FALSE); + _mesa_set_enable(ctx, GL_TEXTURE_GEN_T, GL_FALSE); + _mesa_set_enable(ctx, GL_TEXTURE_GEN_R, GL_FALSE); + _mesa_set_enable(ctx, GL_TEXTURE_GEN_Q, GL_FALSE); + } + } + + /* save current texture objects for unit[0] only */ + for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++) { + _mesa_reference_texobj(&save->CurrentTexture[tgt], + ctx->Texture.Unit[0].CurrentTex[tgt]); + } + + /* set defaults for unit[0] */ + _mesa_ActiveTextureARB(GL_TEXTURE0); + _mesa_ClientActiveTextureARB(GL_TEXTURE0); + _mesa_TexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE); + } + + if (state & META_TRANSFORM) { + GLuint activeTexture = ctx->Texture.CurrentUnit; + _mesa_memcpy(save->ModelviewMatrix, ctx->ModelviewMatrixStack.Top->m, + 16 * sizeof(GLfloat)); + _mesa_memcpy(save->ProjectionMatrix, ctx->ProjectionMatrixStack.Top->m, + 16 * sizeof(GLfloat)); + _mesa_memcpy(save->TextureMatrix, ctx->TextureMatrixStack[0].Top->m, + 16 * sizeof(GLfloat)); + save->MatrixMode = ctx->Transform.MatrixMode; + /* set 1:1 vertex:pixel coordinate transform */ + _mesa_ActiveTextureARB(GL_TEXTURE0); + _mesa_MatrixMode(GL_TEXTURE); + _mesa_LoadIdentity(); + _mesa_ActiveTextureARB(GL_TEXTURE0 + activeTexture); + _mesa_MatrixMode(GL_MODELVIEW); + _mesa_LoadIdentity(); + _mesa_MatrixMode(GL_PROJECTION); + _mesa_LoadIdentity(); + _mesa_Ortho(0.0F, ctx->DrawBuffer->Width, + 0.0F, ctx->DrawBuffer->Height, + -1.0F, 1.0F); + save->ClipPlanesEnabled = ctx->Transform.ClipPlanesEnabled; + if (ctx->Transform.ClipPlanesEnabled) { + GLuint i; + for (i = 0; i < ctx->Const.MaxClipPlanes; i++) { + _mesa_set_enable(ctx, GL_CLIP_PLANE0 + i, GL_FALSE); + } + } + } + + if (state & META_VERTEX) { + /* save vertex array object state */ + _mesa_reference_array_object(ctx, &save->ArrayObj, + ctx->Array.ArrayObj); + _mesa_reference_buffer_object(ctx, &save->ArrayBufferObj, + ctx->Array.ArrayBufferObj); + /* set some default state? */ + } + + if (state & META_VIEWPORT) { + save->ViewportX = ctx->Viewport.X; + save->ViewportY = ctx->Viewport.Y; + save->ViewportW = ctx->Viewport.Width; + save->ViewportH = ctx->Viewport.Height; + _mesa_Viewport(0, 0, ctx->DrawBuffer->Width, ctx->DrawBuffer->Height); + save->DepthNear = ctx->Viewport.Near; + save->DepthFar = ctx->Viewport.Far; + _mesa_DepthRange(0.0, 1.0); + } + + /* misc */ + { + save->Lighting = ctx->Light.Enabled; + if (ctx->Light.Enabled) + _mesa_set_enable(ctx, GL_LIGHTING, GL_FALSE); + } +} + + +/** + * Leave meta state. This is like a light-weight version of glPopAttrib(). + */ +static void +_mesa_meta_end(GLcontext *ctx) +{ + struct save_state *save = &ctx->Meta->Save; + const GLbitfield state = save->SavedState; + + if (state & META_ALPHA_TEST) { + if (ctx->Color.AlphaEnabled != save->AlphaEnabled) + _mesa_set_enable(ctx, GL_ALPHA_TEST, save->AlphaEnabled); + } + + if (state & META_BLEND) { + if (ctx->Color.BlendEnabled != save->BlendEnabled) + _mesa_set_enable(ctx, GL_BLEND, save->BlendEnabled); + if (ctx->Color.ColorLogicOpEnabled != save->ColorLogicOpEnabled) + _mesa_set_enable(ctx, GL_COLOR_LOGIC_OP, save->ColorLogicOpEnabled); + } + + if (state & META_COLOR_MASK) { + if (!TEST_EQ_4V(ctx->Color.ColorMask, save->ColorMask)) + _mesa_ColorMask(save->ColorMask[0], save->ColorMask[1], + save->ColorMask[2], save->ColorMask[3]); + } + + if (state & META_DEPTH_TEST) { + if (ctx->Depth.Test != save->Depth.Test) + _mesa_set_enable(ctx, GL_DEPTH_TEST, save->Depth.Test); + _mesa_DepthFunc(save->Depth.Func); + _mesa_DepthMask(save->Depth.Mask); + } + + if (state & META_FOG) { + _mesa_set_enable(ctx, GL_FOG, save->Fog); + } + + if (state & META_RASTERIZATION) { + _mesa_PolygonMode(GL_FRONT, save->FrontPolygonMode); + _mesa_PolygonMode(GL_BACK, save->BackPolygonMode); + _mesa_set_enable(ctx, GL_POLYGON_STIPPLE, save->PolygonStipple); + _mesa_set_enable(ctx, GL_POLYGON_OFFSET_FILL, save->PolygonOffset); + _mesa_set_enable(ctx, GL_POLYGON_SMOOTH, save->PolygonSmooth); + _mesa_set_enable(ctx, GL_CULL_FACE, save->PolygonCull); + } + + if (state & META_SCISSOR) { + _mesa_set_enable(ctx, GL_SCISSOR_TEST, save->Scissor.Enabled); + _mesa_Scissor(save->Scissor.X, save->Scissor.Y, + save->Scissor.Width, save->Scissor.Height); + } + + if (state & META_SHADER) { + if (ctx->Extensions.ARB_vertex_program) { + _mesa_set_enable(ctx, GL_VERTEX_PROGRAM_ARB, + save->VertexProgramEnabled); + _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current, + save->VertexProgram); + } + + if (ctx->Extensions.ARB_fragment_program) { + _mesa_set_enable(ctx, GL_FRAGMENT_PROGRAM_ARB, + save->FragmentProgramEnabled); + _mesa_reference_fragprog(ctx, &ctx->FragmentProgram.Current, + save->FragmentProgram); + } + + if (ctx->Extensions.ARB_shader_objects) { + _mesa_UseProgramObjectARB(save->Shader); + } + } + + if (state & META_STENCIL_TEST) { + const struct gl_stencil_attrib *stencil = &save->Stencil; + + _mesa_set_enable(ctx, GL_STENCIL_TEST, stencil->Enabled); + _mesa_ClearStencil(stencil->Clear); + if (ctx->Extensions.EXT_stencil_two_side) { + _mesa_set_enable(ctx, GL_STENCIL_TEST_TWO_SIDE_EXT, + stencil->TestTwoSide); + _mesa_ActiveStencilFaceEXT(stencil->ActiveFace + ? GL_BACK : GL_FRONT); + } + /* front state */ + _mesa_StencilFuncSeparate(GL_FRONT, + stencil->Function[0], + stencil->Ref[0], + stencil->ValueMask[0]); + _mesa_StencilMaskSeparate(GL_FRONT, stencil->WriteMask[0]); + _mesa_StencilOpSeparate(GL_FRONT, stencil->FailFunc[0], + stencil->ZFailFunc[0], + stencil->ZPassFunc[0]); + /* back state */ + _mesa_StencilFuncSeparate(GL_BACK, + stencil->Function[1], + stencil->Ref[1], + stencil->ValueMask[1]); + _mesa_StencilMaskSeparate(GL_BACK, stencil->WriteMask[1]); + _mesa_StencilOpSeparate(GL_BACK, stencil->FailFunc[1], + stencil->ZFailFunc[1], + stencil->ZPassFunc[1]); + } + + if (state & META_TEXTURE) { + GLuint u, tgt; + + ASSERT(ctx->Texture.CurrentUnit == 0); + + /* restore texenv for unit[0] */ + _mesa_TexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, save->EnvMode); + + /* restore texture objects for unit[0] only */ + for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++) { + _mesa_reference_texobj(&ctx->Texture.Unit[0].CurrentTex[tgt], + save->CurrentTexture[tgt]); + } + + /* Re-enable textures, texgen */ + for (u = 0; u < ctx->Const.MaxTextureUnits; u++) { + if (save->TexEnabled[u]) { + _mesa_ActiveTextureARB(GL_TEXTURE0 + u); + + if (save->TexEnabled[u] & TEXTURE_1D_BIT) + _mesa_set_enable(ctx, GL_TEXTURE_1D, GL_TRUE); + if (save->TexEnabled[u] & TEXTURE_2D_BIT) + _mesa_set_enable(ctx, GL_TEXTURE_2D, GL_TRUE); + if (save->TexEnabled[u] & TEXTURE_3D_BIT) + _mesa_set_enable(ctx, GL_TEXTURE_3D, GL_TRUE); + if (save->TexEnabled[u] & TEXTURE_CUBE_BIT) + _mesa_set_enable(ctx, GL_TEXTURE_CUBE_MAP, GL_TRUE); + if (save->TexEnabled[u] & TEXTURE_RECT_BIT) + _mesa_set_enable(ctx, GL_TEXTURE_RECTANGLE, GL_TRUE); + } + + if (save->TexGenEnabled[u]) { + _mesa_ActiveTextureARB(GL_TEXTURE0 + u); + + if (save->TexGenEnabled[u] & S_BIT) + _mesa_set_enable(ctx, GL_TEXTURE_GEN_S, GL_TRUE); + if (save->TexGenEnabled[u] & T_BIT) + _mesa_set_enable(ctx, GL_TEXTURE_GEN_T, GL_TRUE); + if (save->TexGenEnabled[u] & R_BIT) + _mesa_set_enable(ctx, GL_TEXTURE_GEN_R, GL_TRUE); + if (save->TexGenEnabled[u] & Q_BIT) + _mesa_set_enable(ctx, GL_TEXTURE_GEN_Q, GL_TRUE); + } + } + + /* restore current unit state */ + _mesa_ActiveTextureARB(GL_TEXTURE0 + save->ActiveUnit); + _mesa_ClientActiveTextureARB(GL_TEXTURE0 + save->ClientActiveUnit); + } + + if (state & META_TRANSFORM) { + GLuint activeTexture = ctx->Texture.CurrentUnit; + _mesa_ActiveTextureARB(GL_TEXTURE0); + _mesa_MatrixMode(GL_TEXTURE); + _mesa_LoadMatrixf(save->TextureMatrix); + _mesa_ActiveTextureARB(GL_TEXTURE0 + activeTexture); + + _mesa_MatrixMode(GL_MODELVIEW); + _mesa_LoadMatrixf(save->ModelviewMatrix); + + _mesa_MatrixMode(GL_PROJECTION); + _mesa_LoadMatrixf(save->ProjectionMatrix); + + _mesa_MatrixMode(save->MatrixMode); + + save->ClipPlanesEnabled = ctx->Transform.ClipPlanesEnabled; + if (save->ClipPlanesEnabled) { + GLuint i; + for (i = 0; i < ctx->Const.MaxClipPlanes; i++) { + if (save->ClipPlanesEnabled & (1 << i)) { + _mesa_set_enable(ctx, GL_CLIP_PLANE0 + i, GL_TRUE); + } + } + } + } + + if (state & META_VERTEX) { + /* restore vertex buffer object */ + _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, save->ArrayBufferObj->Name); + _mesa_reference_buffer_object(ctx, &save->ArrayBufferObj, NULL); + + /* restore vertex array object */ + _mesa_BindVertexArray(save->ArrayObj->Name); + _mesa_reference_array_object(ctx, &save->ArrayObj, NULL); + } + + if (state & META_VIEWPORT) { + _mesa_Viewport(save->ViewportX, save->ViewportY, + save->ViewportW, save->ViewportH); + _mesa_DepthRange(save->DepthNear, save->DepthFar); + } + + /* misc */ + if (save->Lighting) { + _mesa_set_enable(ctx, GL_LIGHTING, GL_TRUE); + } + if (save->Fog) { + _mesa_set_enable(ctx, GL_FOG, GL_TRUE); + } +} + + +/** + * Meta implementation of ctx->Driver.BlitFramebuffer() in terms + * of texture mapping and polygon rendering. + * Note: this function requires GL_ARB_texture_rectangle support. + */ +void +_mesa_meta_blit_framebuffer(GLcontext *ctx, + GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, + GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, + GLbitfield mask, GLenum filter) +{ + struct blit_state *blit = &ctx->Meta->Blit; + const GLint srcX = MIN2(srcX0, srcX1); + const GLint srcY = MIN2(srcY0, srcY1); + const GLint srcW = abs(srcX1 - srcX0); + const GLint srcH = abs(srcY1 - srcY0); + GLboolean srcFlipX = srcX1 < srcX0; + GLboolean srcFlipY = srcY1 < srcY0; + + ASSERT(ctx->Extensions.NV_texture_rectangle); + + if (srcW > ctx->Const.MaxTextureRectSize || + srcH > ctx->Const.MaxTextureRectSize) { + /* XXX avoid this fallback */ + _swrast_BlitFramebuffer(ctx, srcX0, srcY0, srcX1, srcY1, + dstX0, dstY0, dstX1, dstY1, mask, filter); + return; + } + + + if (srcFlipX) { + GLint tmp = dstX0; + dstX0 = dstX1; + dstX1 = tmp; + } + + if (srcFlipY) { + GLint tmp = dstY0; + dstY0 = dstY1; + dstY1 = tmp; + } + + /* only scissor effects blit so save/clear all other relevant state */ + _mesa_meta_begin(ctx, ~META_SCISSOR); + + if (blit->TexObj == 0) { + /* one-time setup */ + + /* create texture object */ + _mesa_GenTextures(1, &blit->TexObj); + _mesa_BindTexture(GL_TEXTURE_RECTANGLE, blit->TexObj); + _mesa_TexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE); + } + else { + _mesa_BindTexture(GL_TEXTURE_RECTANGLE, blit->TexObj); + } + + _mesa_TexParameteri(GL_TEXTURE_RECTANGLE, GL_TEXTURE_MIN_FILTER, filter); + _mesa_TexParameteri(GL_TEXTURE_RECTANGLE, GL_TEXTURE_MAG_FILTER, filter); + + if (blit->ArrayObj == 0) { + /* one-time setup */ + + /* create vertex array object */ + _mesa_GenVertexArrays(1, &blit->ArrayObj); + _mesa_BindVertexArray(blit->ArrayObj); + + /* create vertex array buffer */ + _mesa_GenBuffersARB(1, &blit->VBO); + _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, blit->VBO); + _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(blit->verts), + blit->verts, GL_STREAM_DRAW_ARB); + + /* setup vertex arrays */ + _mesa_VertexPointer(2, GL_FLOAT, 4 * sizeof(GLfloat), + (void*) (0 * sizeof(GLfloat))); + _mesa_TexCoordPointer(2, GL_FLOAT, 4 * sizeof(GLfloat), + (void *) (2 * sizeof(GLfloat))); + _mesa_EnableClientState(GL_VERTEX_ARRAY); + _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); + } + else { + _mesa_BindVertexArray(blit->ArrayObj); + _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, blit->VBO); + } + + /* vertex positions */ + blit->verts[0][0] = (GLfloat) dstX0; + blit->verts[0][1] = (GLfloat) dstY0; + blit->verts[1][0] = (GLfloat) dstX1; + blit->verts[1][1] = (GLfloat) dstY0; + blit->verts[2][0] = (GLfloat) dstX1; + blit->verts[2][1] = (GLfloat) dstY1; + blit->verts[3][0] = (GLfloat) dstX0; + blit->verts[3][1] = (GLfloat) dstY1; + + /* texcoords */ + blit->verts[0][2] = 0.0F; + blit->verts[0][3] = 0.0F; + blit->verts[1][2] = (GLfloat) srcW; + blit->verts[1][3] = 0.0F; + blit->verts[2][2] = (GLfloat) srcW; + blit->verts[2][3] = (GLfloat) srcH; + blit->verts[3][2] = 0.0F; + blit->verts[3][3] = (GLfloat) srcH; + + /* upload new vertex data */ + _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, + sizeof(blit->verts), blit->verts); + + /* copy framebuffer image to texture */ + if (mask & GL_COLOR_BUFFER_BIT) { + if (blit->TexWidth == srcW && + blit->TexHeight == srcH && + blit->TexType == GL_RGBA) { + /* replace existing tex image */ + _mesa_CopyTexSubImage2D(GL_TEXTURE_RECTANGLE, 0, + 0, 0, srcX, srcY, srcW, srcH); + } + else { + /* create new tex image */ + _mesa_CopyTexImage2D(GL_TEXTURE_RECTANGLE, 0, GL_RGBA, + srcX, srcY, srcW, srcH, 0); + blit->TexWidth = srcW; + blit->TexHeight = srcH; + blit->TexType = GL_RGBA; + } + + mask &= ~GL_COLOR_BUFFER_BIT; + } + + _mesa_Enable(GL_TEXTURE_RECTANGLE); + + /* draw textured quad */ + _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4); + + _mesa_Disable(GL_TEXTURE_RECTANGLE); + + _mesa_meta_end(ctx); + + /* XXX, TO-DO: try to handle these cases above! */ + if (mask & (GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT)) { + _swrast_BlitFramebuffer(ctx, srcX0, srcY0, srcX1, srcY1, + dstX0, dstY0, dstX1, dstY1, mask, filter); + } +} + + +/** + * Meta implementation of ctx->Driver.Clear() in terms of polygon rendering. + */ +void +_mesa_meta_clear(GLcontext *ctx, GLbitfield buffers) +{ + struct clear_state *clear = &ctx->Meta->Clear; + GLfloat z = 1.0 - 2.0 * ctx->Depth.Clear; + GLuint i; + + /* only scissor and color mask effects clearing */ + _mesa_meta_begin(ctx, ~(META_SCISSOR | META_COLOR_MASK)); + + if (clear->ArrayObj == 0) { + /* one-time setup */ + + /* create vertex array object */ + _mesa_GenVertexArrays(1, &clear->ArrayObj); + _mesa_BindVertexArray(clear->ArrayObj); + + /* create vertex array buffer */ + _mesa_GenBuffersARB(1, &clear->VBO); + _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, clear->VBO); + _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(clear->verts), + clear->verts, GL_STREAM_DRAW_ARB); + + /* setup vertex arrays */ + _mesa_VertexPointer(3, GL_FLOAT, 7 * sizeof(GLfloat), (void *) 0); + _mesa_ColorPointer(4, GL_FLOAT, 7 * sizeof(GLfloat), + (void *) (3 * sizeof(GLfloat))); + _mesa_EnableClientState(GL_VERTEX_ARRAY); + _mesa_EnableClientState(GL_COLOR_ARRAY); + } + else { + _mesa_BindVertexArray(clear->ArrayObj); + _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, clear->VBO); + } + + /* GL_COLOR_BUFFER_BIT */ + if (buffers & BUFFER_BITS_COLOR) { + /* leave colormask, glDrawBuffer state as-is */ + } + else { + _mesa_ColorMask(GL_FALSE, GL_FALSE, GL_FALSE, GL_FALSE); + } + + /* GL_DEPTH_BUFFER_BIT */ + if (buffers & BUFFER_BIT_DEPTH) { + _mesa_set_enable(ctx, GL_DEPTH_TEST, GL_TRUE); + _mesa_DepthFunc(GL_ALWAYS); + _mesa_DepthMask(GL_TRUE); + } + else { + assert(!ctx->Depth.Test); + } + + /* GL_STENCIL_BUFFER_BIT */ + if (buffers & BUFFER_BIT_STENCIL) { + _mesa_set_enable(ctx, GL_STENCIL_TEST, GL_TRUE); + _mesa_StencilOpSeparate(GL_FRONT_AND_BACK, + GL_REPLACE, GL_REPLACE, GL_REPLACE); + _mesa_StencilFuncSeparate(GL_FRONT_AND_BACK, GL_ALWAYS, + ctx->Stencil.Clear & 0x7fffffff, + ctx->Stencil.WriteMask[0]); + } + else { + assert(!ctx->Stencil.Enabled); + } + + /* vertex positions */ + clear->verts[0][0] = (GLfloat) ctx->DrawBuffer->_Xmin; + clear->verts[0][1] = (GLfloat) ctx->DrawBuffer->_Ymin; + clear->verts[0][2] = z; + clear->verts[1][0] = (GLfloat) ctx->DrawBuffer->_Xmax; + clear->verts[1][1] = (GLfloat) ctx->DrawBuffer->_Ymin; + clear->verts[1][2] = z; + clear->verts[2][0] = (GLfloat) ctx->DrawBuffer->_Xmax; + clear->verts[2][1] = (GLfloat) ctx->DrawBuffer->_Ymax; + clear->verts[2][2] = z; + clear->verts[3][0] = (GLfloat) ctx->DrawBuffer->_Xmin; + clear->verts[3][1] = (GLfloat) ctx->DrawBuffer->_Ymax; + clear->verts[3][2] = z; + + /* vertex colors */ + for (i = 0; i < 4; i++) { + COPY_4FV(&clear->verts[i][3], ctx->Color.ClearColor); + } + + /* upload new vertex data */ + _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, + sizeof(clear->verts), clear->verts); + + /* draw quad */ + _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4); + + _mesa_meta_end(ctx); +} + + +/** + * Meta implementation of ctx->Driver.CopyPixels() in terms + * of texture mapping and polygon rendering. + * Note: this function requires GL_ARB_texture_rectangle support. + */ +void +_mesa_meta_copy_pixels(GLcontext *ctx, GLint srcX, GLint srcY, + GLsizei width, GLsizei height, + GLint dstX, GLint dstY, GLenum type) +{ + const GLenum filter = GL_NEAREST; + struct copypix_state *copypix = &ctx->Meta->CopyPix; + const GLfloat z = ctx->Current.RasterPos[2]; + const GLfloat dstX1 = dstX + width * ctx->Pixel.ZoomX; + const GLfloat dstY1 = dstY + height * ctx->Pixel.ZoomY; + + ASSERT(ctx->Extensions.NV_texture_rectangle); + + if (type != GL_COLOR || + ctx->_ImageTransferState || + ctx->Fog.Enabled || + width > ctx->Const.MaxTextureRectSize || + height > ctx->Const.MaxTextureRectSize) { + /* XXX avoid this fallback */ + _swrast_CopyPixels(ctx, srcX, srcY, width, height, dstX, dstY, type); + return; + } + + /* Most GL state applies to glCopyPixels, but a there's a few things + * we need to override: + */ + _mesa_meta_begin(ctx, (META_RASTERIZATION | + META_SHADER | + META_TEXTURE | + META_TRANSFORM | + META_VERTEX | + META_VIEWPORT)); + + if (copypix->TexObj == 0) { + /* one-time setup */ + + /* create texture object */ + _mesa_GenTextures(1, ©pix->TexObj); + _mesa_BindTexture(GL_TEXTURE_RECTANGLE, copypix->TexObj); + _mesa_TexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE); + _mesa_TexParameteri(GL_TEXTURE_RECTANGLE, GL_TEXTURE_MIN_FILTER, filter); + _mesa_TexParameteri(GL_TEXTURE_RECTANGLE, GL_TEXTURE_MAG_FILTER, filter); + } + else { + _mesa_BindTexture(GL_TEXTURE_RECTANGLE, copypix->TexObj); + } + + if (copypix->ArrayObj == 0) { + /* one-time setup */ + + /* create vertex array object */ + _mesa_GenVertexArrays(1, ©pix->ArrayObj); + _mesa_BindVertexArray(copypix->ArrayObj); + + /* create vertex array buffer */ + _mesa_GenBuffersARB(1, ©pix->VBO); + _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, copypix->VBO); + _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(copypix->verts), + copypix->verts, GL_STREAM_DRAW_ARB); + + /* setup vertex arrays */ + _mesa_VertexPointer(3, GL_FLOAT, sizeof(copypix->verts[0]), + (void*) (0 * sizeof(GLfloat))); + _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(copypix->verts[0]), + (void *) (3 * sizeof(GLfloat))); + _mesa_EnableClientState(GL_VERTEX_ARRAY); + _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); + } + else { + _mesa_BindVertexArray(copypix->ArrayObj); + _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, copypix->VBO); + } + + /* vertex positions, texcoords */ + copypix->verts[0][0] = (GLfloat) dstX; + copypix->verts[0][1] = (GLfloat) dstY; + copypix->verts[0][2] = z; + copypix->verts[0][3] = 0.0F; + copypix->verts[0][4] = 0.0F; + copypix->verts[1][0] = (GLfloat) dstX1; + copypix->verts[1][1] = (GLfloat) dstY; + copypix->verts[1][2] = z; + copypix->verts[1][3] = (GLfloat) width; + copypix->verts[1][4] = 0.0F; + copypix->verts[2][0] = (GLfloat) dstX1; + copypix->verts[2][1] = (GLfloat) dstY1; + copypix->verts[2][2] = z; + copypix->verts[2][3] = (GLfloat) width; + copypix->verts[2][4] = (GLfloat) height; + copypix->verts[3][0] = (GLfloat) dstX; + copypix->verts[3][1] = (GLfloat) dstY1; + copypix->verts[3][2] = z; + copypix->verts[3][3] = 0.0F; + copypix->verts[3][4] = (GLfloat) height; + + /* upload new vertex data */ + _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, + sizeof(copypix->verts), copypix->verts); + + /* copy framebuffer image to texture */ + if (type == GL_COLOR) { + if (copypix->TexWidth == width && + copypix->TexHeight == height && + copypix->TexType == type) { + /* replace existing tex image */ + _mesa_CopyTexSubImage2D(GL_TEXTURE_RECTANGLE, 0, + 0, 0, srcX, srcY, width, height); + } + else { + /* create new tex image */ + _mesa_CopyTexImage2D(GL_TEXTURE_RECTANGLE, 0, GL_RGBA, + srcX, srcY, width, height, 0); + copypix->TexWidth = width; + copypix->TexHeight = height; + copypix->TexType = type; + } + } + else if (type == GL_DEPTH) { + /* TO-DO: Use a GL_DEPTH_COMPONENT texture and a fragment program/shader + * that replaces the fragment.z value. + */ + } + else { + ASSERT(type == GL_STENCIL); + /* have to use sw fallback */ + } + + _mesa_Enable(GL_TEXTURE_RECTANGLE); + + /* draw textured quad */ + _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4); + + _mesa_Disable(GL_TEXTURE_RECTANGLE); + + _mesa_meta_end(ctx); +} + + + +/** + * When the glDrawPixels() image size is greater than the max rectangle + * texture size we use this function to break the glDrawPixels() image + * into tiles which fit into the max texture size. + */ +static void +tiled_draw_pixels(GLcontext *ctx, + GLint x, GLint y, GLsizei width, GLsizei height, + GLenum format, GLenum type, + const struct gl_pixelstore_attrib *unpack, + const GLvoid *pixels) +{ + const GLint maxSize = ctx->Const.MaxTextureRectSize; + struct gl_pixelstore_attrib tileUnpack = *unpack; + GLint i, j; + + for (i = 0; i < width; i += maxSize) { + const GLint tileWidth = MIN2(maxSize, width - i); + const GLint tileX = (GLint) (x + i * ctx->Pixel.ZoomX); + + tileUnpack.SkipPixels = unpack->SkipPixels + i; + + for (j = 0; j < height; j += maxSize) { + const GLint tileHeight = MIN2(maxSize, height - j); + const GLint tileY = (GLint) (y + j * ctx->Pixel.ZoomY); + + tileUnpack.SkipRows = unpack->SkipRows + j; + + _mesa_meta_draw_pixels(ctx, tileX, tileY, + tileWidth, tileHeight, + format, type, &tileUnpack, pixels); + } + } +} + + +/** + * Meta implementation of ctx->Driver.DrawPixels() in terms + * of texture mapping and polygon rendering. + * Note: this function requires GL_ARB_texture_rectangle support. + */ +void +_mesa_meta_draw_pixels(GLcontext *ctx, + GLint x, GLint y, GLsizei width, GLsizei height, + GLenum format, GLenum type, + const struct gl_pixelstore_attrib *unpack, + const GLvoid *pixels) +{ + const GLenum filter = GL_NEAREST; + struct drawpix_state *drawpix = &ctx->Meta->DrawPix; + const GLfloat z = ctx->Current.RasterPos[2]; + const GLfloat x1 = x + width * ctx->Pixel.ZoomX; + const GLfloat y1 = y + height * ctx->Pixel.ZoomY; + const struct gl_pixelstore_attrib unpackSave = ctx->Unpack; + GLenum texIntFormat; + GLboolean fallback; + + ASSERT(ctx->Extensions.NV_texture_rectangle); + + /* + * Determine if we can do the glDrawPixels with texture mapping. + */ + fallback = GL_FALSE; + if (ctx->_ImageTransferState || + ctx->Fog.Enabled) { + fallback = GL_TRUE; + } + + if (_mesa_is_color_format(format)) { + texIntFormat = GL_RGBA; + } + else { + fallback = GL_TRUE; + } + + if (fallback) { + _swrast_DrawPixels(ctx, x, y, width, height, + format, type, unpack, pixels); + return; + } + + /* + * Check image size against max texture size, draw as tiles if needed. + */ + if (width > ctx->Const.MaxTextureRectSize || + height > ctx->Const.MaxTextureRectSize) { + tiled_draw_pixels(ctx, x, y, width, height, + format, type, unpack, pixels); + return; + } + + /* Most GL state applies to glDrawPixels, but a there's a few things + * we need to override: + */ + _mesa_meta_begin(ctx, (META_RASTERIZATION | + META_SHADER | + META_TEXTURE | + META_TRANSFORM | + META_VERTEX | + META_VIEWPORT)); + + if (drawpix->TexObj == 0) { + /* one-time setup */ + + /* create texture object */ + _mesa_GenTextures(1, &drawpix->TexObj); + _mesa_BindTexture(GL_TEXTURE_RECTANGLE, drawpix->TexObj); + _mesa_TexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE); + _mesa_TexParameteri(GL_TEXTURE_RECTANGLE, GL_TEXTURE_MIN_FILTER, filter); + _mesa_TexParameteri(GL_TEXTURE_RECTANGLE, GL_TEXTURE_MAG_FILTER, filter); + } + else { + _mesa_BindTexture(GL_TEXTURE_RECTANGLE, drawpix->TexObj); + } + + if (drawpix->ArrayObj == 0) { + /* one-time setup */ + + /* create vertex array object */ + _mesa_GenVertexArrays(1, &drawpix->ArrayObj); + _mesa_BindVertexArray(drawpix->ArrayObj); + + /* create vertex array buffer */ + _mesa_GenBuffersARB(1, &drawpix->VBO); + _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, drawpix->VBO); + _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(drawpix->verts), + drawpix->verts, GL_STREAM_DRAW_ARB); + + /* setup vertex arrays */ + _mesa_VertexPointer(3, GL_FLOAT, sizeof(drawpix->verts[0]), + (void*) (0 * sizeof(GLfloat))); + _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(drawpix->verts[0]), + (void *) (3 * sizeof(GLfloat))); + _mesa_EnableClientState(GL_VERTEX_ARRAY); + _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); + } + else { + _mesa_BindVertexArray(drawpix->ArrayObj); + _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, drawpix->VBO); + } + + /* vertex positions, texcoords */ + drawpix->verts[0][0] = (GLfloat) x; + drawpix->verts[0][1] = (GLfloat) y; + drawpix->verts[0][2] = z; + drawpix->verts[0][3] = 0.0F; + drawpix->verts[0][4] = 0.0F; + drawpix->verts[1][0] = (GLfloat) x1; + drawpix->verts[1][1] = (GLfloat) y; + drawpix->verts[1][2] = z; + drawpix->verts[1][3] = (GLfloat) width; + drawpix->verts[1][4] = 0.0F; + drawpix->verts[2][0] = (GLfloat) x1; + drawpix->verts[2][1] = (GLfloat) y1; + drawpix->verts[2][2] = z; + drawpix->verts[2][3] = (GLfloat) width; + drawpix->verts[2][4] = (GLfloat) height; + drawpix->verts[3][0] = (GLfloat) x; + drawpix->verts[3][1] = (GLfloat) y1; + drawpix->verts[3][2] = z; + drawpix->verts[3][3] = 0.0F; + drawpix->verts[3][4] = (GLfloat) height; + + /* upload new vertex data */ + _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, + sizeof(drawpix->verts), drawpix->verts); + + /* set given unpack params */ + ctx->Unpack = *unpack; /* XXX bufobj */ + + /* copy pixel data to texture */ + if (drawpix->TexWidth == width && + drawpix->TexHeight == height && + drawpix->TexIntFormat == texIntFormat) { + /* replace existing tex image */ + _mesa_TexSubImage2D(GL_TEXTURE_RECTANGLE, 0, + 0, 0, width, height, format, type, pixels); + } + else { + /* create new tex image */ + _mesa_TexImage2D(GL_TEXTURE_RECTANGLE, 0, texIntFormat, + width, height, 0, format, type, pixels); + drawpix->TexWidth = width; + drawpix->TexHeight = height; + drawpix->TexIntFormat = texIntFormat; + } + + /* restore unpack params */ + ctx->Unpack = unpackSave; /* XXX bufobj */ + + _mesa_Enable(GL_TEXTURE_RECTANGLE); + + /* draw textured quad */ + _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4); + + _mesa_Disable(GL_TEXTURE_RECTANGLE); + + _mesa_meta_end(ctx); +} diff --git a/src/mesa/drivers/common/meta.h b/src/mesa/drivers/common/meta.h new file mode 100644 index 00000000000..a9c5f980432 --- /dev/null +++ b/src/mesa/drivers/common/meta.h @@ -0,0 +1,80 @@ +/* + * Mesa 3-D graphics library + * Version: 7.6 + * + * Copyright (C) 2009 VMware, Inc. All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + + +#ifndef META_H +#define META_H + + +/** + * Flags passed to _mesa_meta_begin(). + * XXX these flags may evolve... + */ +/*@{*/ +#define META_ALPHA_TEST 0x1 +#define META_BLEND 0x2 /**< includes logicop */ +#define META_COLOR_MASK 0x4 +#define META_DEPTH_TEST 0x8 +#define META_FOG 0x10 +#define META_RASTERIZATION 0x20 +#define META_SCISSOR 0x40 +#define META_SHADER 0x80 +#define META_STENCIL_TEST 0x100 +#define META_TRANSFORM 0x200 /**< modelview, projection */ +#define META_TEXTURE 0x400 +#define META_VERTEX 0x800 +#define META_VIEWPORT 0x1000 +#define META_ALL ~0x0 +/*@}*/ + + +extern void +_mesa_meta_init(GLcontext *ctx); + +extern void +_mesa_meta_free(GLcontext *ctx); + +extern void +_mesa_meta_blit_framebuffer(GLcontext *ctx, + GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, + GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, + GLbitfield mask, GLenum filter); + +extern void +_mesa_meta_clear(GLcontext *ctx, GLbitfield buffers); + +extern void +_mesa_meta_copy_pixels(GLcontext *ctx, GLint srcx, GLint srcy, + GLsizei width, GLsizei height, + GLint dstx, GLint dsty, GLenum type); + +extern void +_mesa_meta_draw_pixels(GLcontext *ctx, + GLint x, GLint y, GLsizei width, GLsizei height, + GLenum format, GLenum type, + const struct gl_pixelstore_attrib *unpack, + const GLvoid *pixels); + + +#endif /* META_H */ diff --git a/src/mesa/drivers/dri/common/.gitignore b/src/mesa/drivers/dri/common/.gitignore new file mode 100644 index 00000000000..1edeb79fd12 --- /dev/null +++ b/src/mesa/drivers/dri/common/.gitignore @@ -0,0 +1 @@ +*.os diff --git a/src/mesa/drivers/dri/common/dri_metaops.c b/src/mesa/drivers/dri/common/dri_metaops.c index fe183c2e870..cdbea344951 100644 --- a/src/mesa/drivers/dri/common/dri_metaops.c +++ b/src/mesa/drivers/dri/common/dri_metaops.c @@ -357,6 +357,7 @@ meta_clear_tris(struct dri_metaops *meta, GLbitfield mask) GLuint saved_shader_program = 0; unsigned int saved_active_texture; struct gl_array_object *arraySave = NULL; + GLfloat saved_near, saved_far; if (!meta->clear.arrayObj) meta_init_clear(meta); @@ -370,8 +371,7 @@ meta_clear_tris(struct dri_metaops *meta, GLbitfield mask) GL_POLYGON_BIT | GL_STENCIL_BUFFER_BIT | GL_TRANSFORM_BIT | - GL_CURRENT_BIT | - GL_VIEWPORT_BIT); + GL_CURRENT_BIT); saved_active_texture = ctx->Texture.CurrentUnit; /* Disable existing GL state we don't want to apply to a clear. */ @@ -440,6 +440,8 @@ meta_clear_tris(struct dri_metaops *meta, GLbitfield mask) /* The ClearDepth value is unaffected by DepthRange, so do a default * mapping. */ + saved_near = ctx->Viewport.Near; + saved_far = ctx->Viewport.Far; _mesa_DepthRange(0.0, 1.0); /* Prepare the vertices, which are the same regardless of which buffer we're @@ -519,6 +521,7 @@ meta_clear_tris(struct dri_metaops *meta, GLbitfield mask) if (saved_shader_program) _mesa_UseProgramObjectARB(saved_shader_program); + _mesa_DepthRange(saved_near, saved_far); _mesa_PopAttrib(); /* restore current array object */ diff --git a/src/mesa/drivers/dri/common/dri_util.c b/src/mesa/drivers/dri/common/dri_util.c index 1d940603fa8..e48e10d7c06 100644 --- a/src/mesa/drivers/dri/common/dri_util.c +++ b/src/mesa/drivers/dri/common/dri_util.c @@ -778,7 +778,7 @@ dri2CreateNewScreen(int scrn, int fd, if (driDriverAPI.InitScreen2 == NULL) return NULL; - psp = _mesa_malloc(sizeof(*psp)); + psp = _mesa_calloc(sizeof(*psp)); if (!psp) return NULL; diff --git a/src/mesa/drivers/dri/common/extension_helper.h b/src/mesa/drivers/dri/common/extension_helper.h index 62801494ce3..08a97bb1110 100644 --- a/src/mesa/drivers/dri/common/extension_helper.h +++ b/src/mesa/drivers/dri/common/extension_helper.h @@ -752,6 +752,13 @@ static const char VertexAttrib4ubNV_names[] = ""; #endif +#if defined(need_GL_APPLE_texture_range) +static const char TextureRangeAPPLE_names[] = + "iip\0" /* Parameter signature */ + "glTextureRangeAPPLE\0" + ""; +#endif + #if defined(need_GL_SUN_vertex) static const char TexCoord2fColor4fNormal3fVertex3fSUN_names[] = "ffffffffffff\0" /* Parameter signature */ @@ -1567,6 +1574,13 @@ static const char FragmentLightivSGIX_names[] = ""; #endif +#if defined(need_GL_APPLE_texture_range) +static const char GetTexParameterPointervAPPLE_names[] = + "iip\0" /* Parameter signature */ + "glGetTexParameterPointervAPPLE\0" + ""; +#endif + #if defined(need_GL_EXT_pixel_transform) static const char PixelTransformParameterfvEXT_names[] = "iip\0" /* Parameter signature */ @@ -1771,6 +1785,13 @@ static const char DeleteFencesNV_names[] = ""; #endif +#if defined(need_GL_SGIX_polynomial_ffd) +static const char DeformationMap3dSGIX_names[] = + "iddiiddiiddiip\0" /* Parameter signature */ + "glDeformationMap3dSGIX\0" + ""; +#endif + #if defined(need_GL_VERSION_2_0) static const char IsShader_names[] = "i\0" /* Parameter signature */ @@ -1997,13 +2018,6 @@ static const char GetCombinerOutputParameterivNV_names[] = ""; #endif -#if defined(need_GL_IBM_multimode_draw_arrays) -static const char MultiModeDrawArraysIBM_names[] = - "pppii\0" /* Parameter signature */ - "glMultiModeDrawArraysIBM\0" - ""; -#endif - #if defined(need_GL_SGIS_pixel_texture) static const char PixelTexGenParameterivSGIS_names[] = "ip\0" /* Parameter signature */ @@ -2079,6 +2093,13 @@ static const char Uniform2fvARB_names[] = ""; #endif +#if defined(need_GL_APPLE_flush_buffer_range) +static const char BufferParameteriAPPLE_names[] = + "iii\0" /* Parameter signature */ + "glBufferParameteriAPPLE\0" + ""; +#endif + #if defined(need_GL_VERSION_1_3) static const char MultiTexCoord3dvARB_names[] = "ip\0" /* Parameter signature */ @@ -2803,6 +2824,13 @@ static const char IsProgramNV_names[] = ""; #endif +#if defined(need_GL_APPLE_flush_buffer_range) +static const char FlushMappedBufferRangeAPPLE_names[] = + "iii\0" /* Parameter signature */ + "glFlushMappedBufferRangeAPPLE\0" + ""; +#endif + #if defined(need_GL_SUN_triangle_list) static const char ReplacementCodePointerSUN_names[] = "iip\0" /* Parameter signature */ @@ -3920,6 +3948,13 @@ static const char VertexAttribs4dvNV_names[] = ""; #endif +#if defined(need_GL_IBM_multimode_draw_arrays) +static const char MultiModeDrawArraysIBM_names[] = + "pppii\0" /* Parameter signature */ + "glMultiModeDrawArraysIBM\0" + ""; +#endif + #if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program) static const char VertexAttrib4dARB_names[] = "idddd\0" /* Parameter signature */ @@ -4604,13 +4639,6 @@ static const char Minmax_names[] = ""; #endif -#if defined(need_GL_SGIX_polynomial_ffd) -static const char DeformationMap3dSGIX_names[] = - "iddiiddiiddiip\0" /* Parameter signature */ - "glDeformationMap3dSGIX\0" - ""; -#endif - #if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_fog_coord) static const char FogCoorddvEXT_names[] = "p\0" /* Parameter signature */ @@ -4973,6 +5001,22 @@ static const struct dri_extension_function GL_3DFX_tbuffer_functions[] = { }; #endif +#if defined(need_GL_APPLE_flush_buffer_range) +static const struct dri_extension_function GL_APPLE_flush_buffer_range_functions[] = { + { BufferParameteriAPPLE_names, BufferParameteriAPPLE_remap_index, -1 }, + { FlushMappedBufferRangeAPPLE_names, FlushMappedBufferRangeAPPLE_remap_index, -1 }, + { NULL, 0, 0 } +}; +#endif + +#if defined(need_GL_APPLE_texture_range) +static const struct dri_extension_function GL_APPLE_texture_range_functions[] = { + { TextureRangeAPPLE_names, TextureRangeAPPLE_remap_index, -1 }, + { GetTexParameterPointervAPPLE_names, GetTexParameterPointervAPPLE_remap_index, -1 }, + { NULL, 0, 0 } +}; +#endif + #if defined(need_GL_APPLE_vertex_array_object) static const struct dri_extension_function GL_APPLE_vertex_array_object_functions[] = { { DeleteVertexArraysAPPLE_names, DeleteVertexArraysAPPLE_remap_index, -1 }, @@ -6131,9 +6175,9 @@ static const struct dri_extension_function GL_SGIX_pixel_texture_functions[] = { #if defined(need_GL_SGIX_polynomial_ffd) static const struct dri_extension_function GL_SGIX_polynomial_ffd_functions[] = { { LoadIdentityDeformationMapSGIX_names, LoadIdentityDeformationMapSGIX_remap_index, -1 }, + { DeformationMap3dSGIX_names, DeformationMap3dSGIX_remap_index, -1 }, { DeformSGIX_names, DeformSGIX_remap_index, -1 }, { DeformationMap3fSGIX_names, DeformationMap3fSGIX_remap_index, -1 }, - { DeformationMap3dSGIX_names, DeformationMap3dSGIX_remap_index, -1 }, { NULL, 0, 0 } }; #endif diff --git a/src/mesa/drivers/dri/fb/fb_egl.c b/src/mesa/drivers/dri/fb/fb_egl.c index dee67feb5ac..4e41860d8c8 100644 --- a/src/mesa/drivers/dri/fb/fb_egl.c +++ b/src/mesa/drivers/dri/fb/fb_egl.c @@ -605,7 +605,7 @@ fbDestroySurface(_EGLDriver *drv, EGLDisplay dpy, EGLSurface surface) { fbSurface *fs = Lookup_fbSurface(surface); _eglUnlinkSurface(&fs->Base); - if (!fs->Base.IsBound) + if (!_eglIsSurfaceBound(&fs->Base)) free(fs); return EGL_TRUE; } @@ -616,7 +616,7 @@ fbDestroyContext(_EGLDriver *drv, EGLDisplay dpy, EGLContext context) { fbContext *fc = Lookup_fbContext(context); _eglUnlinkContext(&fc->Base); - if (!fc->Base.IsBound) + if (!_eglIsContextBound(&fc->Base)) free(fc); return EGL_TRUE; } diff --git a/src/mesa/drivers/dri/i915/i830_context.h b/src/mesa/drivers/dri/i915/i830_context.h index 1bdb32049d7..f73cbbf88bb 100644 --- a/src/mesa/drivers/dri/i915/i830_context.h +++ b/src/mesa/drivers/dri/i915/i830_context.h @@ -40,6 +40,7 @@ #define I830_UPLOAD_BUFFERS 0x2 #define I830_UPLOAD_STIPPLE 0x4 #define I830_UPLOAD_INVARIENT 0x8 +#define I830_UPLOAD_RASTER_RULES 0x10 #define I830_UPLOAD_TEX(i) (0x10<<(i)) #define I830_UPLOAD_TEXBLEND(i) (0x100<<(i)) #define I830_UPLOAD_TEX_ALL (0x0f0) @@ -99,6 +100,11 @@ #define I830_TEXBLEND_SIZE 12 /* (4 args + op) * 2 + COLOR_FACTOR */ +enum { + I830_RASTER_RULES, + I830_RASTER_RULES_SIZE +}; + struct i830_texture_object { struct intel_texture_object intel; @@ -112,6 +118,7 @@ struct i830_hw_state GLuint Ctx[I830_CTX_SETUP_SIZE]; GLuint Buffer[I830_DEST_SETUP_SIZE]; GLuint Stipple[I830_STP_SETUP_SIZE]; + GLuint RasterRules[I830_RASTER_RULES_SIZE]; GLuint Tex[I830_TEX_UNITS][I830_TEX_SETUP_SIZE]; GLuint TexBlend[I830_TEX_UNITS][I830_TEXBLEND_SIZE]; GLuint TexBlendWordsUsed[I830_TEX_UNITS]; @@ -197,6 +204,7 @@ extern void i830InitStateFuncs(struct dd_function_table *functions); extern void i830EmitState(struct i830_context *i830); extern void i830InitState(struct i830_context *i830); +extern void i830_update_provoking_vertex(GLcontext *ctx); /* i830_metaops.c */ diff --git a/src/mesa/drivers/dri/i915/i830_reg.h b/src/mesa/drivers/dri/i915/i830_reg.h index db16871001d..ae1317029a2 100644 --- a/src/mesa/drivers/dri/i915/i830_reg.h +++ b/src/mesa/drivers/dri/i915/i830_reg.h @@ -420,8 +420,11 @@ #define ENABLE_LINE_STRIP_PROVOKE_VRTX (1<<8) #define ENABLE_TRI_FAN_PROVOKE_VRTX (1<<5) #define ENABLE_TRI_STRIP_PROVOKE_VRTX (1<<2) +#define LINE_STRIP_PROVOKE_VRTX_MASK (3<<6) #define LINE_STRIP_PROVOKE_VRTX(x) ((x)<<6) +#define TRI_FAN_PROVOKE_VRTX_MASK (3<<3) #define TRI_FAN_PROVOKE_VRTX(x) ((x)<<3) +#define TRI_STRIP_PROVOKE_VRTX_MASK (3<<0) #define TRI_STRIP_PROVOKE_VRTX(x) (x) /* _3DSTATE_SCISSOR_ENABLE, p200 */ diff --git a/src/mesa/drivers/dri/i915/i830_state.c b/src/mesa/drivers/dri/i915/i830_state.c index 8ef6c9144f1..645ebe30577 100644 --- a/src/mesa/drivers/dri/i915/i830_state.c +++ b/src/mesa/drivers/dri/i915/i830_state.c @@ -1047,6 +1047,16 @@ i830_init_packets(struct i830_context *i830) TEXBIND_SET1(TEXCOORDSRC_VTXSET_1) | TEXBIND_SET0(TEXCOORDSRC_VTXSET_0)); + i830->state.RasterRules[I830_RASTER_RULES] = (_3DSTATE_RASTER_RULES_CMD | + ENABLE_POINT_RASTER_RULE | + OGL_POINT_RASTER_RULE | + ENABLE_LINE_STRIP_PROVOKE_VRTX | + ENABLE_TRI_FAN_PROVOKE_VRTX | + ENABLE_TRI_STRIP_PROVOKE_VRTX | + LINE_STRIP_PROVOKE_VRTX(1) | + TRI_FAN_PROVOKE_VRTX(2) | + TRI_STRIP_PROVOKE_VRTX(2)); + i830->state.Stipple[I830_STPREG_ST0] = _3DSTATE_STIPPLE; @@ -1058,6 +1068,27 @@ i830_init_packets(struct i830_context *i830) i830->state.Buffer[I830_DESTREG_SR2] = 0; } +void +i830_update_provoking_vertex(GLcontext * ctx) +{ + struct i830_context *i830 = i830_context(ctx); + + I830_STATECHANGE(i830, I830_UPLOAD_RASTER_RULES); + i830->state.RasterRules[I830_RASTER_RULES] &= ~(LINE_STRIP_PROVOKE_VRTX_MASK | + TRI_FAN_PROVOKE_VRTX_MASK | + TRI_STRIP_PROVOKE_VRTX_MASK); + + /* _NEW_LIGHT */ + if (ctx->Light.ProvokingVertex == GL_LAST_VERTEX_CONVENTION) { + i830->state.RasterRules[I830_RASTER_RULES] |= (LINE_STRIP_PROVOKE_VRTX(1) | + TRI_FAN_PROVOKE_VRTX(2) | + TRI_STRIP_PROVOKE_VRTX(2)); + } else { + i830->state.RasterRules[I830_RASTER_RULES] |= (LINE_STRIP_PROVOKE_VRTX(0) | + TRI_FAN_PROVOKE_VRTX(1) | + TRI_STRIP_PROVOKE_VRTX(0)); + } +} void i830InitStateFuncs(struct dd_function_table *functions) @@ -1101,6 +1132,7 @@ i830InitState(struct i830_context *i830) i830->current = &i830->state; i830->state.emitted = 0; i830->state.active = (I830_UPLOAD_INVARIENT | + I830_UPLOAD_RASTER_RULES | I830_UPLOAD_TEXBLEND(0) | I830_UPLOAD_STIPPLE | I830_UPLOAD_CTX | I830_UPLOAD_BUFFERS); diff --git a/src/mesa/drivers/dri/i915/i830_vtbl.c b/src/mesa/drivers/dri/i915/i830_vtbl.c index 9c6f891dd32..983f6724c98 100644 --- a/src/mesa/drivers/dri/i915/i830_vtbl.c +++ b/src/mesa/drivers/dri/i915/i830_vtbl.c @@ -299,7 +299,7 @@ i830_emit_invarient_state(struct intel_context *intel) { BATCH_LOCALS; - BEGIN_BATCH(30, IGNORE_CLIPRECTS); + BEGIN_BATCH(29, IGNORE_CLIPRECTS); OUT_BATCH(_3DSTATE_DFLT_DIFFUSE_CMD); OUT_BATCH(0); @@ -351,15 +351,6 @@ i830_emit_invarient_state(struct intel_context *intel) OUT_BATCH(_3DSTATE_MAP_COORD_TRANSFORM); OUT_BATCH(DISABLE_TEX_TRANSFORM | TEXTURE_SET(3)); - OUT_BATCH(_3DSTATE_RASTER_RULES_CMD | - ENABLE_POINT_RASTER_RULE | - OGL_POINT_RASTER_RULE | - ENABLE_LINE_STRIP_PROVOKE_VRTX | - ENABLE_TRI_FAN_PROVOKE_VRTX | - ENABLE_TRI_STRIP_PROVOKE_VRTX | - LINE_STRIP_PROVOKE_VRTX(1) | - TRI_FAN_PROVOKE_VRTX(2) | TRI_STRIP_PROVOKE_VRTX(2)); - OUT_BATCH(_3DSTATE_VERTEX_TRANSFORM); OUT_BATCH(DISABLE_VIEWPORT_TRANSFORM | DISABLE_PERSPECTIVE_DIVIDE); @@ -394,6 +385,9 @@ get_state_size(struct i830_hw_state *state) if (dirty & I830_UPLOAD_INVARIENT) sz += 40 * sizeof(int); + if (dirty & I830_UPLOAD_RASTER_RULES) + sz += sizeof(state->RasterRules); + if (dirty & I830_UPLOAD_CTX) sz += sizeof(state->Ctx); @@ -486,6 +480,11 @@ i830_emit_state(struct intel_context *intel) i830_emit_invarient_state(intel); } + if (dirty & I830_UPLOAD_RASTER_RULES) { + DBG("I830_UPLOAD_RASTER_RULES:\n"); + emit(intel, state->RasterRules, sizeof(state->RasterRules)); + } + if (dirty & I830_UPLOAD_CTX) { DBG("I830_UPLOAD_CTX:\n"); emit(intel, state->Ctx, sizeof(state->Ctx)); @@ -737,6 +736,13 @@ i830_assert_not_dirty( struct intel_context *intel ) assert(!get_dirty(state)); } +static void +i830_invalidate_state(struct intel_context *intel, GLuint new_state) +{ + if (new_state & _NEW_LIGHT) + i830_update_provoking_vertex(&intel->ctx); +} + void i830InitVtbl(struct i830_context *i830) { @@ -752,4 +758,5 @@ i830InitVtbl(struct i830_context *i830) i830->intel.vtbl.render_prevalidate = i830_render_prevalidate; i830->intel.vtbl.assert_not_dirty = i830_assert_not_dirty; i830->intel.vtbl.finish_batch = intel_finish_vb; + i830->intel.vtbl.invalidate_state = i830_invalidate_state; } diff --git a/src/mesa/drivers/dri/i915/i915_context.c b/src/mesa/drivers/dri/i915/i915_context.c index 367d2a3b648..bb08cf8d18f 100644 --- a/src/mesa/drivers/dri/i915/i915_context.c +++ b/src/mesa/drivers/dri/i915/i915_context.c @@ -75,6 +75,10 @@ i915InvalidateState(GLcontext * ctx, GLuint new_state) if (new_state & (_NEW_FOG | _NEW_HINT | _NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS)) i915_update_fog(ctx); + if (new_state & (_NEW_STENCIL | _NEW_BUFFERS | _NEW_POLYGON)) + i915_update_stencil(ctx); + if (new_state & (_NEW_LIGHT)) + i915_update_provoking_vertex(ctx); } diff --git a/src/mesa/drivers/dri/i915/i915_context.h b/src/mesa/drivers/dri/i915/i915_context.h index 87bbf5f9271..8de4a9d0d36 100644 --- a/src/mesa/drivers/dri/i915/i915_context.h +++ b/src/mesa/drivers/dri/i915/i915_context.h @@ -48,6 +48,7 @@ #define I915_UPLOAD_FOG 0x20 #define I915_UPLOAD_INVARIENT 0x40 #define I915_UPLOAD_DEFAULTS 0x80 +#define I915_UPLOAD_RASTER_RULES 0x100 #define I915_UPLOAD_TEX(i) (0x00010000<<(i)) #define I915_UPLOAD_TEX_ALL (0x00ff0000) #define I915_UPLOAD_TEX_0_SHIFT 16 @@ -82,7 +83,9 @@ #define I915_CTXREG_IAB 6 #define I915_CTXREG_BLENDCOLOR0 7 #define I915_CTXREG_BLENDCOLOR1 8 -#define I915_CTX_SETUP_SIZE 9 +#define I915_CTXREG_BF_STENCIL_OPS 9 +#define I915_CTXREG_BF_STENCIL_MASKS 10 +#define I915_CTX_SETUP_SIZE 11 #define I915_FOGREG_COLOR 0 #define I915_FOGREG_MODE0 1 @@ -110,6 +113,10 @@ #define I915_DEFREG_Z1 5 #define I915_DEF_SETUP_SIZE 6 +enum { + I915_RASTER_RULES, + I915_RASTER_RULES_SETUP_SIZE, +}; #define I915_MAX_CONSTANT 32 #define I915_CONSTANT_SIZE (2+(4*I915_MAX_CONSTANT)) @@ -206,6 +213,7 @@ struct i915_hw_state GLuint Stipple[I915_STP_SETUP_SIZE]; GLuint Fog[I915_FOG_SETUP_SIZE]; GLuint Defaults[I915_DEF_SETUP_SIZE]; + GLuint RasterRules[I915_RASTER_RULES_SETUP_SIZE]; GLuint Tex[I915_TEX_UNITS][I915_TEX_SETUP_SIZE]; GLuint Constant[I915_CONSTANT_SIZE]; GLuint ConstantSize; @@ -321,6 +329,8 @@ extern void i915_print_ureg(const char *msg, GLuint ureg); extern void i915InitStateFunctions(struct dd_function_table *functions); extern void i915InitState(struct i915_context *i915); extern void i915_update_fog(GLcontext * ctx); +extern void i915_update_stencil(GLcontext * ctx); +extern void i915_update_provoking_vertex(GLcontext *ctx); /*====================================================================== diff --git a/src/mesa/drivers/dri/i915/i915_reg.h b/src/mesa/drivers/dri/i915/i915_reg.h index 84db58ea950..b5fa7fddb96 100644 --- a/src/mesa/drivers/dri/i915/i915_reg.h +++ b/src/mesa/drivers/dri/i915/i915_reg.h @@ -86,8 +86,10 @@ #define BFM_ENABLE_STENCIL_WRITE_MASK (1<<16) #define BFM_STENCIL_TEST_MASK_SHIFT 8 #define BFM_STENCIL_TEST_MASK_MASK (0xff<<8) +#define BFM_STENCIL_TEST_MASK(x) (((x)&0xff) << 8) #define BFM_STENCIL_WRITE_MASK_SHIFT 0 #define BFM_STENCIL_WRITE_MASK_MASK (0xff<<0) +#define BFM_STENCIL_WRITE_MASK(x) ((x)&0xff) @@ -295,7 +297,9 @@ #define TEXKILL_4D (1<<9) #define ENABLE_LINE_STRIP_PROVOKE_VRTX (1<<8) #define ENABLE_TRI_FAN_PROVOKE_VRTX (1<<5) +#define LINE_STRIP_PROVOKE_VRTX_MASK (3 << 6) #define LINE_STRIP_PROVOKE_VRTX(x) ((x)<<6) +#define TRI_FAN_PROVOKE_VRTX_MASK (3 << 3) #define TRI_FAN_PROVOKE_VRTX(x) ((x)<<3) /* _3DSTATE_SCISSOR_ENABLE, p256 */ diff --git a/src/mesa/drivers/dri/i915/i915_state.c b/src/mesa/drivers/dri/i915/i915_state.c index 814fb59fd34..b60efea75bd 100644 --- a/src/mesa/drivers/dri/i915/i915_state.c +++ b/src/mesa/drivers/dri/i915/i915_state.c @@ -48,73 +48,119 @@ #define FILE_DEBUG_FLAG DEBUG_STATE -static void -i915StencilFuncSeparate(GLcontext * ctx, GLenum face, GLenum func, GLint ref, - GLuint mask) +void +i915_update_stencil(GLcontext * ctx) { struct i915_context *i915 = I915_CONTEXT(ctx); - int test = intel_translate_compare_func(func); + GLuint front_ref, front_writemask, front_mask; + GLenum front_func, front_fail, front_pass_z_fail, front_pass_z_pass; + GLuint back_ref, back_writemask, back_mask; + GLenum back_func, back_fail, back_pass_z_fail, back_pass_z_pass; - mask = mask & 0xff; - - DBG("%s : func: %s, ref : 0x%x, mask: 0x%x\n", __FUNCTION__, - _mesa_lookup_enum_by_nr(func), ref, mask); + I915_STATECHANGE(i915, I915_UPLOAD_CTX); + /* The 915 considers CW to be "front" for two-sided stencil, so choose + * appropriately. + */ + /* _NEW_POLYGON | _NEW_STENCIL */ + if (ctx->Polygon.FrontFace == GL_CW) { + front_ref = ctx->Stencil.Ref[0]; + front_mask = ctx->Stencil.ValueMask[0]; + front_writemask = ctx->Stencil.WriteMask[0]; + front_func = ctx->Stencil.Function[0]; + front_fail = ctx->Stencil.FailFunc[0]; + front_pass_z_fail = ctx->Stencil.ZFailFunc[0]; + front_pass_z_pass = ctx->Stencil.ZPassFunc[0]; + back_ref = ctx->Stencil.Ref[ctx->Stencil._BackFace]; + back_mask = ctx->Stencil.ValueMask[ctx->Stencil._BackFace]; + back_writemask = ctx->Stencil.WriteMask[ctx->Stencil._BackFace]; + back_func = ctx->Stencil.Function[ctx->Stencil._BackFace]; + back_fail = ctx->Stencil.FailFunc[ctx->Stencil._BackFace]; + back_pass_z_fail = ctx->Stencil.ZFailFunc[ctx->Stencil._BackFace]; + back_pass_z_pass = ctx->Stencil.ZPassFunc[ctx->Stencil._BackFace]; + } else { + front_ref = ctx->Stencil.Ref[ctx->Stencil._BackFace]; + front_mask = ctx->Stencil.ValueMask[ctx->Stencil._BackFace]; + front_writemask = ctx->Stencil.WriteMask[ctx->Stencil._BackFace]; + front_func = ctx->Stencil.Function[ctx->Stencil._BackFace]; + front_fail = ctx->Stencil.FailFunc[ctx->Stencil._BackFace]; + front_pass_z_fail = ctx->Stencil.ZFailFunc[ctx->Stencil._BackFace]; + front_pass_z_pass = ctx->Stencil.ZPassFunc[ctx->Stencil._BackFace]; + back_ref = ctx->Stencil.Ref[0]; + back_mask = ctx->Stencil.ValueMask[0]; + back_writemask = ctx->Stencil.WriteMask[0]; + back_func = ctx->Stencil.Function[0]; + back_fail = ctx->Stencil.FailFunc[0]; + back_pass_z_fail = ctx->Stencil.ZFailFunc[0]; + back_pass_z_pass = ctx->Stencil.ZPassFunc[0]; + } - I915_STATECHANGE(i915, I915_UPLOAD_CTX); - i915->state.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_TEST_MASK; + /* Set front state. */ + i915->state.Ctx[I915_CTXREG_STATE4] &= ~(MODE4_ENABLE_STENCIL_TEST_MASK | + MODE4_ENABLE_STENCIL_WRITE_MASK); i915->state.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_TEST_MASK | - STENCIL_TEST_MASK(mask)); + ENABLE_STENCIL_WRITE_MASK | + STENCIL_TEST_MASK(front_mask) | + STENCIL_WRITE_MASK(front_writemask)); i915->state.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_REF_MASK | - S5_STENCIL_TEST_FUNC_MASK); + S5_STENCIL_TEST_FUNC_MASK | + S5_STENCIL_FAIL_MASK | + S5_STENCIL_PASS_Z_FAIL_MASK | + S5_STENCIL_PASS_Z_PASS_MASK); + + i915->state.Ctx[I915_CTXREG_LIS5] |= + (front_ref << S5_STENCIL_REF_SHIFT) | + (intel_translate_compare_func(front_func) << S5_STENCIL_TEST_FUNC_SHIFT) | + (intel_translate_stencil_op(front_fail) << S5_STENCIL_FAIL_SHIFT) | + (intel_translate_stencil_op(front_pass_z_fail) << + S5_STENCIL_PASS_Z_FAIL_SHIFT) | + (intel_translate_stencil_op(front_pass_z_pass) << + S5_STENCIL_PASS_Z_PASS_SHIFT); + + /* Set back state if different from front. */ + if (ctx->Stencil._TestTwoSide) { + i915->state.Ctx[I915_CTXREG_BF_STENCIL_OPS] &= + ~(BFO_STENCIL_REF_MASK | + BFO_STENCIL_TEST_MASK | + BFO_STENCIL_FAIL_MASK | + BFO_STENCIL_PASS_Z_FAIL_MASK | + BFO_STENCIL_PASS_Z_PASS_MASK); + i915->state.Ctx[I915_CTXREG_BF_STENCIL_OPS] |= BFO_STENCIL_TWO_SIDE | + (back_ref << BFO_STENCIL_REF_SHIFT) | + (intel_translate_compare_func(back_func) << BFO_STENCIL_TEST_SHIFT) | + (intel_translate_stencil_op(back_fail) << BFO_STENCIL_FAIL_SHIFT) | + (intel_translate_stencil_op(back_pass_z_fail) << + BFO_STENCIL_PASS_Z_FAIL_SHIFT) | + (intel_translate_stencil_op(back_pass_z_pass) << + BFO_STENCIL_PASS_Z_PASS_SHIFT); + + i915->state.Ctx[I915_CTXREG_BF_STENCIL_MASKS] &= + ~(BFM_STENCIL_TEST_MASK_MASK | + BFM_STENCIL_WRITE_MASK_MASK); + i915->state.Ctx[I915_CTXREG_BF_STENCIL_MASKS] |= + BFM_STENCIL_TEST_MASK(back_mask) | + BFM_STENCIL_WRITE_MASK(back_writemask); + } else { + i915->state.Ctx[I915_CTXREG_BF_STENCIL_OPS] &= ~BFO_STENCIL_TWO_SIDE; + } +} - i915->state.Ctx[I915_CTXREG_LIS5] |= ((ref << S5_STENCIL_REF_SHIFT) | - (test << - S5_STENCIL_TEST_FUNC_SHIFT)); +static void +i915StencilFuncSeparate(GLcontext * ctx, GLenum face, GLenum func, GLint ref, + GLuint mask) +{ } static void i915StencilMaskSeparate(GLcontext * ctx, GLenum face, GLuint mask) { - struct i915_context *i915 = I915_CONTEXT(ctx); - - DBG("%s : mask 0x%x\n", __FUNCTION__, mask); - - mask = mask & 0xff; - - I915_STATECHANGE(i915, I915_UPLOAD_CTX); - i915->state.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_WRITE_MASK; - i915->state.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_WRITE_MASK | - STENCIL_WRITE_MASK(mask)); } - static void i915StencilOpSeparate(GLcontext * ctx, GLenum face, GLenum fail, GLenum zfail, GLenum zpass) { - struct i915_context *i915 = I915_CONTEXT(ctx); - int fop = intel_translate_stencil_op(fail); - int dfop = intel_translate_stencil_op(zfail); - int dpop = intel_translate_stencil_op(zpass); - - - DBG("%s: fail : %s, zfail: %s, zpass : %s\n", __FUNCTION__, - _mesa_lookup_enum_by_nr(fail), - _mesa_lookup_enum_by_nr(zfail), _mesa_lookup_enum_by_nr(zpass)); - - I915_STATECHANGE(i915, I915_UPLOAD_CTX); - - i915->state.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_FAIL_MASK | - S5_STENCIL_PASS_Z_FAIL_MASK | - S5_STENCIL_PASS_Z_PASS_MASK); - - i915->state.Ctx[I915_CTXREG_LIS5] |= ((fop << S5_STENCIL_FAIL_SHIFT) | - (dfop << - S5_STENCIL_PASS_Z_FAIL_SHIFT) | - (dpop << - S5_STENCIL_PASS_Z_PASS_SHIFT)); } static void @@ -945,6 +991,17 @@ i915_init_packets(struct i915_context *i915) _3DSTATE_CONST_BLEND_COLOR_CMD; i915->state.Ctx[I915_CTXREG_BLENDCOLOR1] = 0; + i915->state.Ctx[I915_CTXREG_BF_STENCIL_MASKS] = + _3DSTATE_BACKFACE_STENCIL_MASKS | + BFM_ENABLE_STENCIL_TEST_MASK | + BFM_ENABLE_STENCIL_WRITE_MASK | + (0xff << BFM_STENCIL_WRITE_MASK_SHIFT) | + (0xff << BFM_STENCIL_TEST_MASK_SHIFT); + i915->state.Ctx[I915_CTXREG_BF_STENCIL_OPS] = + _3DSTATE_BACKFACE_STENCIL_OPS | + BFO_ENABLE_STENCIL_REF | + BFO_ENABLE_STENCIL_FUNCS | + BFO_ENABLE_STENCIL_TWO_SIDE; } { @@ -976,6 +1033,13 @@ i915_init_packets(struct i915_context *i915) i915->state.Buffer[I915_DESTREG_SR2] = 0; } + i915->state.RasterRules[I915_RASTER_RULES] = _3DSTATE_RASTER_RULES_CMD | + ENABLE_POINT_RASTER_RULE | + OGL_POINT_RASTER_RULE | + ENABLE_LINE_STRIP_PROVOKE_VRTX | + ENABLE_TRI_FAN_PROVOKE_VRTX | + LINE_STRIP_PROVOKE_VRTX(1) | + TRI_FAN_PROVOKE_VRTX(2) | ENABLE_TEXKILL_3D_4D | TEXKILL_4D; #if 0 { @@ -996,7 +1060,33 @@ i915_init_packets(struct i915_context *i915) i915->state.active = (I915_UPLOAD_PROGRAM | I915_UPLOAD_STIPPLE | I915_UPLOAD_CTX | - I915_UPLOAD_BUFFERS | I915_UPLOAD_INVARIENT); + I915_UPLOAD_BUFFERS | + I915_UPLOAD_INVARIENT | + I915_UPLOAD_RASTER_RULES); +} + +void +i915_update_provoking_vertex(GLcontext * ctx) +{ + struct i915_context *i915 = I915_CONTEXT(ctx); + + I915_STATECHANGE(i915, I915_UPLOAD_CTX); + i915->state.Ctx[I915_CTXREG_LIS6] &= ~(S6_TRISTRIP_PV_MASK); + + I915_STATECHANGE(i915, I915_UPLOAD_RASTER_RULES); + i915->state.RasterRules[I915_RASTER_RULES] &= ~(LINE_STRIP_PROVOKE_VRTX_MASK | + TRI_FAN_PROVOKE_VRTX_MASK); + + /* _NEW_LIGHT */ + if (ctx->Light.ProvokingVertex == GL_LAST_VERTEX_CONVENTION) { + i915->state.RasterRules[I915_RASTER_RULES] |= (LINE_STRIP_PROVOKE_VRTX(1) | + TRI_FAN_PROVOKE_VRTX(2)); + i915->state.Ctx[I915_CTXREG_LIS6] |= (2 << S6_TRISTRIP_PV_SHIFT); + } else { + i915->state.RasterRules[I915_RASTER_RULES] |= (LINE_STRIP_PROVOKE_VRTX(0) | + TRI_FAN_PROVOKE_VRTX(1)); + i915->state.Ctx[I915_CTXREG_LIS6] |= (0 << S6_TRISTRIP_PV_SHIFT); + } } void diff --git a/src/mesa/drivers/dri/i915/i915_vtbl.c b/src/mesa/drivers/dri/i915/i915_vtbl.c index fe1be93a6d0..9a723d3cd73 100644 --- a/src/mesa/drivers/dri/i915/i915_vtbl.c +++ b/src/mesa/drivers/dri/i915/i915_vtbl.c @@ -176,7 +176,7 @@ i915_emit_invarient_state(struct intel_context *intel) { BATCH_LOCALS; - BEGIN_BATCH(20, IGNORE_CLIPRECTS); + BEGIN_BATCH(17, IGNORE_CLIPRECTS); OUT_BATCH(_3DSTATE_AA_CMD | AA_LINE_ECAAR_WIDTH_ENABLE | @@ -200,14 +200,6 @@ i915_emit_invarient_state(struct intel_context *intel) CSB_TCB(3, 3) | CSB_TCB(4, 4) | CSB_TCB(5, 5) | CSB_TCB(6, 6) | CSB_TCB(7, 7)); - OUT_BATCH(_3DSTATE_RASTER_RULES_CMD | - ENABLE_POINT_RASTER_RULE | - OGL_POINT_RASTER_RULE | - ENABLE_LINE_STRIP_PROVOKE_VRTX | - ENABLE_TRI_FAN_PROVOKE_VRTX | - LINE_STRIP_PROVOKE_VRTX(1) | - TRI_FAN_PROVOKE_VRTX(2) | ENABLE_TEXKILL_3D_4D | TEXKILL_4D); - /* Need to initialize this to zero. */ OUT_BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 | I1_LOAD_S(3) | (0)); @@ -225,11 +217,6 @@ i915_emit_invarient_state(struct intel_context *intel) OUT_BATCH(_3DSTATE_LOAD_INDIRECT | 0); /* disable indirect state */ OUT_BATCH(0); - - /* Don't support twosided stencil yet */ - OUT_BATCH(_3DSTATE_BACKFACE_STENCIL_OPS | BFO_ENABLE_STENCIL_TWO_SIDE | 0); - OUT_BATCH(0); - ADVANCE_BATCH(); } @@ -263,6 +250,9 @@ get_state_size(struct i915_hw_state *state) if (dirty & I915_UPLOAD_INVARIENT) sz += 30 * 4; + if (dirty & I915_UPLOAD_RASTER_RULES) + sz += sizeof(state->RasterRules); + if (dirty & I915_UPLOAD_CTX) sz += sizeof(state->Ctx); @@ -371,6 +361,12 @@ i915_emit_state(struct intel_context *intel) i915_emit_invarient_state(intel); } + if (dirty & I915_UPLOAD_RASTER_RULES) { + if (INTEL_DEBUG & DEBUG_STATE) + fprintf(stderr, "I915_UPLOAD_RASTER_RULES:\n"); + emit(intel, state->RasterRules, sizeof(state->RasterRules)); + } + if (dirty & I915_UPLOAD_CTX) { if (INTEL_DEBUG & DEBUG_STATE) fprintf(stderr, "I915_UPLOAD_CTX:\n"); diff --git a/src/mesa/drivers/dri/i965/Makefile b/src/mesa/drivers/dri/i965/Makefile index 00a42111da0..128afb56866 100644 --- a/src/mesa/drivers/dri/i965/Makefile +++ b/src/mesa/drivers/dri/i965/Makefile @@ -43,6 +43,7 @@ DRIVER_SOURCES = \ brw_clip_util.c \ brw_context.c \ brw_curbe.c \ + brw_disasm.c \ brw_draw.c \ brw_draw_upload.c \ brw_eu.c \ diff --git a/src/mesa/drivers/dri/i965/brw_clip.h b/src/mesa/drivers/dri/i965/brw_clip.h index 12e8548df1f..957df441ab0 100644 --- a/src/mesa/drivers/dri/i965/brw_clip.h +++ b/src/mesa/drivers/dri/i965/brw_clip.h @@ -100,6 +100,8 @@ struct brw_clip_compile { struct brw_reg fixed_planes; struct brw_reg plane_equation; + + struct brw_reg ff_sync; } reg; /* 3 different ways of expressing vertex size: @@ -173,4 +175,5 @@ struct brw_reg get_tmp( struct brw_clip_compile *c ); void brw_clip_project_position(struct brw_clip_compile *c, struct brw_reg pos ); void brw_clip_ff_sync(struct brw_clip_compile *c); +void brw_clip_init_ff_sync(struct brw_clip_compile *c); #endif diff --git a/src/mesa/drivers/dri/i965/brw_clip_line.c b/src/mesa/drivers/dri/i965/brw_clip_line.c index 9abd0642aa2..048ca620fab 100644 --- a/src/mesa/drivers/dri/i965/brw_clip_line.c +++ b/src/mesa/drivers/dri/i965/brw_clip_line.c @@ -85,6 +85,10 @@ static void brw_clip_line_alloc_regs( struct brw_clip_compile *c ) i++; } + if (c->need_ff_sync) { + c->reg.ff_sync = retype(brw_vec1_grf(i, 0), BRW_REGISTER_TYPE_UD); + i++; + } c->first_tmp = i; c->last_tmp = i; @@ -246,8 +250,6 @@ static void clip_and_emit_line( struct brw_clip_compile *c ) brw_ADD(p, c->reg.t, c->reg.t0, c->reg.t1); brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_L, c->reg.t, brw_imm_f(1.0)); - if (c->need_ff_sync) - brw_clip_ff_sync(c); not_culled = brw_IF(p, BRW_EXECUTE_1); { brw_clip_interp_vertex(c, newvtx0, vtx0, vtx1, c->reg.t0, GL_FALSE); @@ -265,6 +267,7 @@ static void clip_and_emit_line( struct brw_clip_compile *c ) void brw_emit_line_clip( struct brw_clip_compile *c ) { brw_clip_line_alloc_regs(c); + brw_clip_init_ff_sync(c); if (c->key.do_flat_shading) brw_clip_copy_colors(c, 0, 1); diff --git a/src/mesa/drivers/dri/i965/brw_clip_point.c b/src/mesa/drivers/dri/i965/brw_clip_point.c index 97382991686..8458f61c5a0 100644 --- a/src/mesa/drivers/dri/i965/brw_clip_point.c +++ b/src/mesa/drivers/dri/i965/brw_clip_point.c @@ -50,7 +50,7 @@ void brw_emit_point_clip( struct brw_clip_compile *c ) /* Send an empty message to kill the thread: */ brw_clip_tri_alloc_regs(c, 0); - if (c->need_ff_sync) - brw_clip_ff_sync(c); + brw_clip_init_ff_sync(c); + brw_clip_kill_thread(c); } diff --git a/src/mesa/drivers/dri/i965/brw_clip_tri.c b/src/mesa/drivers/dri/i965/brw_clip_tri.c index 4c2d655fb10..0efd77225e2 100644 --- a/src/mesa/drivers/dri/i965/brw_clip_tri.c +++ b/src/mesa/drivers/dri/i965/brw_clip_tri.c @@ -119,6 +119,11 @@ void brw_clip_tri_alloc_regs( struct brw_clip_compile *c, i++; } + if (c->need_ff_sync) { + c->reg.ff_sync = retype(brw_vec1_grf(i, 0), BRW_REGISTER_TYPE_UD); + i++; + } + c->first_tmp = i; c->last_tmp = i; @@ -563,6 +568,7 @@ void brw_emit_tri_clip( struct brw_clip_compile *c ) brw_clip_tri_alloc_regs(c, 3 + c->key.nr_userclip + 6); brw_clip_tri_init_vertices(c); brw_clip_init_clipmask(c); + brw_clip_init_ff_sync(c); /* if -ve rhw workaround bit is set, do cliptest */ @@ -589,8 +595,6 @@ void brw_emit_tri_clip( struct brw_clip_compile *c ) else maybe_do_clip_tri(c); - if (c->need_ff_sync) - brw_clip_ff_sync(c); brw_clip_tri_emit_polygon(c); /* Send an empty message to kill the thread: diff --git a/src/mesa/drivers/dri/i965/brw_clip_unfilled.c b/src/mesa/drivers/dri/i965/brw_clip_unfilled.c index 26950383c1b..ad1bfa435fb 100644 --- a/src/mesa/drivers/dri/i965/brw_clip_unfilled.c +++ b/src/mesa/drivers/dri/i965/brw_clip_unfilled.c @@ -453,6 +453,7 @@ void brw_emit_unfilled_clip( struct brw_clip_compile *c ) brw_clip_tri_alloc_regs(c, 3 + c->key.nr_userclip + 6); brw_clip_tri_init_vertices(c); + brw_clip_init_ff_sync(c); assert(c->offset[VERT_RESULT_EDGE]); @@ -496,8 +497,6 @@ void brw_emit_unfilled_clip( struct brw_clip_compile *c ) } brw_ENDIF(p, do_clip); - if (c->need_ff_sync) - brw_clip_ff_sync(c); emit_unfilled_primitives(c); brw_clip_kill_thread(c); } diff --git a/src/mesa/drivers/dri/i965/brw_clip_util.c b/src/mesa/drivers/dri/i965/brw_clip_util.c index e09efc07ed4..5a73abdfee9 100644 --- a/src/mesa/drivers/dri/i965/brw_clip_util.c +++ b/src/mesa/drivers/dri/i965/brw_clip_util.c @@ -213,6 +213,8 @@ void brw_clip_emit_vue(struct brw_clip_compile *c, struct brw_compile *p = &c->func; GLuint start = c->last_mrf; + brw_clip_ff_sync(c); + assert(!(allocate && eot)); /* Cycle through mrf regs - probably futile as we have to wait for @@ -263,6 +265,7 @@ void brw_clip_kill_thread(struct brw_clip_compile *c) { struct brw_compile *p = &c->func; + brw_clip_ff_sync(c); /* Send an empty message to kill the thread and release any * allocated urb entry: */ @@ -356,17 +359,38 @@ void brw_clip_init_clipmask( struct brw_clip_compile *c ) void brw_clip_ff_sync(struct brw_clip_compile *c) { + if (c->need_ff_sync) { + struct brw_compile *p = &c->func; + struct brw_instruction *need_ff_sync; + + brw_set_conditionalmod(p, BRW_CONDITIONAL_Z); + brw_AND(p, brw_null_reg(), c->reg.ff_sync, brw_imm_ud(0x1)); + need_ff_sync = brw_IF(p, BRW_EXECUTE_1); + { + brw_OR(p, c->reg.ff_sync, c->reg.ff_sync, brw_imm_ud(0x1)); + brw_ff_sync(p, + c->reg.R0, + 0, + c->reg.R0, + 1, + 1, /* used */ + 1, /* msg length */ + 1, /* response length */ + 0, /* eot */ + 1, /* write compelete */ + 0, /* urb offset */ + BRW_URB_SWIZZLE_NONE); + } + brw_ENDIF(p, need_ff_sync); + brw_set_predicate_control(p, BRW_PREDICATE_NONE); + } +} + +void brw_clip_init_ff_sync(struct brw_clip_compile *c) +{ + if (c->need_ff_sync) { struct brw_compile *p = &c->func; - brw_ff_sync(p, - c->reg.R0, - 0, - c->reg.R0, - 1, - 1, /* used */ - 1, /* msg length */ - 1, /* response length */ - 0, /* eot */ - 1, /* write compelete */ - 0, /* urb offset */ - BRW_URB_SWIZZLE_NONE); + + brw_MOV(p, c->reg.ff_sync, brw_imm_ud(0)); + } } diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h index 57ddf75413a..847c44ed83a 100644 --- a/src/mesa/drivers/dri/i965/brw_context.h +++ b/src/mesa/drivers/dri/i965/brw_context.h @@ -143,6 +143,7 @@ struct brw_context; #define BRW_NEW_DEPTH_BUFFER 0x20000 #define BRW_NEW_NR_WM_SURFACES 0x40000 #define BRW_NEW_NR_VS_SURFACES 0x80000 +#define BRW_NEW_INDEX_BUFFER 0x100000 struct brw_state_flags { /** State update flags signalled by mesa internals */ @@ -438,9 +439,13 @@ struct brw_query_object { unsigned int count; }; + +/** + * brw_context is derived from intel_context. + */ struct brw_context { - struct intel_context intel; + struct intel_context intel; /**< base class, must be first field */ GLuint primitive; GLboolean emit_state_always; @@ -475,6 +480,9 @@ struct brw_context struct { struct brw_vertex_element inputs[VERT_ATTRIB_MAX]; + struct brw_vertex_element *enabled[VERT_ATTRIB_MAX]; + GLuint nr_enabled; + #define BRW_NR_UPLOAD_BUFS 17 #define BRW_UPLOAD_INIT_SIZE (128*1024) @@ -498,8 +506,15 @@ struct brw_context */ const struct _mesa_index_buffer *ib; + /* Updates to these fields are signaled by BRW_NEW_INDEX_BUFFER. */ dri_bo *bo; unsigned int offset; + unsigned int size; + /* Offset to index buffer index to use in CMD_3D_PRIM so that we can + * avoid re-uploading the IB packet over and over if we're actually + * referencing the same index buffer. + */ + unsigned int start_vertex_offset; } ib; /* Active vertex program: @@ -706,6 +721,8 @@ void brw_upload_urb_fence(struct brw_context *brw); */ void brw_upload_cs_urb_state(struct brw_context *brw); +/* brw_disasm.c */ +int brw_disasm (FILE *file, struct brw_instruction *inst); /*====================================================================== * Inline conversion functions. These are better-typed than the diff --git a/src/mesa/drivers/dri/i965/brw_defines.h b/src/mesa/drivers/dri/i965/brw_defines.h index d166250b4fe..78d457ad2be 100644 --- a/src/mesa/drivers/dri/i965/brw_defines.h +++ b/src/mesa/drivers/dri/i965/brw_defines.h @@ -471,8 +471,9 @@ #define BRW_CONDITIONAL_GE 4 #define BRW_CONDITIONAL_L 5 #define BRW_CONDITIONAL_LE 6 -#define BRW_CONDITIONAL_C 7 +#define BRW_CONDITIONAL_R 7 #define BRW_CONDITIONAL_O 8 +#define BRW_CONDITIONAL_U 9 #define BRW_DEBUG_NONE 0 #define BRW_DEBUG_BREAKPOINT 1 @@ -512,6 +513,7 @@ #define BRW_OPCODE_RSL 11 #define BRW_OPCODE_ASR 12 #define BRW_OPCODE_CMP 16 +#define BRW_OPCODE_CMPN 17 #define BRW_OPCODE_JMPI 32 #define BRW_OPCODE_IF 34 #define BRW_OPCODE_IFF 35 diff --git a/src/mesa/drivers/dri/i965/brw_disasm.c b/src/mesa/drivers/dri/i965/brw_disasm.c new file mode 100644 index 00000000000..9fef230507f --- /dev/null +++ b/src/mesa/drivers/dri/i965/brw_disasm.c @@ -0,0 +1,903 @@ +/* + * Copyright © 2008 Keith Packard + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that copyright + * notice and this permission notice appear in supporting documentation, and + * that the name of the copyright holders not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. The copyright holders make no representations + * about the suitability of this software for any purpose. It is provided "as + * is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, + * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO + * EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY SPECIAL, INDIRECT OR + * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, + * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER + * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE + * OF THIS SOFTWARE. + */ + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <getopt.h> +#include <unistd.h> +#include <stdarg.h> + +#include "main/mtypes.h" + +#include "brw_context.h" +#include "brw_defines.h" + +struct { + char *name; + int nsrc; + int ndst; +} opcode[128] = { + [BRW_OPCODE_MOV] = { .name = "mov", .nsrc = 1, .ndst = 1 }, + [BRW_OPCODE_FRC] = { .name = "frc", .nsrc = 1, .ndst = 1 }, + [BRW_OPCODE_RNDU] = { .name = "rndu", .nsrc = 1, .ndst = 1 }, + [BRW_OPCODE_RNDD] = { .name = "rndd", .nsrc = 1, .ndst = 1 }, + [BRW_OPCODE_RNDE] = { .name = "rnde", .nsrc = 1, .ndst = 1 }, + [BRW_OPCODE_RNDZ] = { .name = "rndz", .nsrc = 1, .ndst = 1 }, + [BRW_OPCODE_NOT] = { .name = "not", .nsrc = 1, .ndst = 1 }, + [BRW_OPCODE_LZD] = { .name = "lzd", .nsrc = 1, .ndst = 1 }, + + [BRW_OPCODE_MUL] = { .name = "mul", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_MAC] = { .name = "mac", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_MACH] = { .name = "mach", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_LINE] = { .name = "line", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_SAD2] = { .name = "sad2", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_SADA2] = { .name = "sada2", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_DP4] = { .name = "dp4", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_DPH] = { .name = "dph", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_DP3] = { .name = "dp3", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_DP2] = { .name = "dp2", .nsrc = 2, .ndst = 1 }, + + [BRW_OPCODE_AVG] = { .name = "avg", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_ADD] = { .name = "add", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_SEL] = { .name = "sel", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_AND] = { .name = "and", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_OR] = { .name = "or", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_XOR] = { .name = "xor", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_SHR] = { .name = "shr", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_SHL] = { .name = "shl", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_ASR] = { .name = "asr", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_CMP] = { .name = "cmp", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_CMPN] = { .name = "cmpn", .nsrc = 2, .ndst = 1 }, + + [BRW_OPCODE_SEND] = { .name = "send", .nsrc = 1, .ndst = 1 }, + [BRW_OPCODE_NOP] = { .name = "nop", .nsrc = 0, .ndst = 0 }, + [BRW_OPCODE_JMPI] = { .name = "jmpi", .nsrc = 1, .ndst = 0 }, + [BRW_OPCODE_IF] = { .name = "if", .nsrc = 2, .ndst = 0 }, + [BRW_OPCODE_IFF] = { .name = "iff", .nsrc = 1, .ndst = 01 }, + [BRW_OPCODE_WHILE] = { .name = "while", .nsrc = 1, .ndst = 0 }, + [BRW_OPCODE_ELSE] = { .name = "else", .nsrc = 2, .ndst = 0 }, + [BRW_OPCODE_BREAK] = { .name = "break", .nsrc = 1, .ndst = 0 }, + [BRW_OPCODE_CONTINUE] = { .name = "cont", .nsrc = 1, .ndst = 0 }, + [BRW_OPCODE_HALT] = { .name = "halt", .nsrc = 1, .ndst = 0 }, + [BRW_OPCODE_MSAVE] = { .name = "msave", .nsrc = 1, .ndst = 1 }, + [BRW_OPCODE_PUSH] = { .name = "push", .nsrc = 1, .ndst = 1 }, + [BRW_OPCODE_MRESTORE] = { .name = "mrest", .nsrc = 1, .ndst = 1 }, + [BRW_OPCODE_POP] = { .name = "pop", .nsrc = 2, .ndst = 0 }, + [BRW_OPCODE_WAIT] = { .name = "wait", .nsrc = 1, .ndst = 0 }, + [BRW_OPCODE_DO] = { .name = "do", .nsrc = 0, .ndst = 0 }, + [BRW_OPCODE_ENDIF] = { .name = "endif", .nsrc = 2, .ndst = 0 }, +}; + +char *conditional_modifier[16] = { + [BRW_CONDITIONAL_NONE] = "", + [BRW_CONDITIONAL_Z] = ".e", + [BRW_CONDITIONAL_NZ] = ".ne", + [BRW_CONDITIONAL_G] = ".g", + [BRW_CONDITIONAL_GE] = ".ge", + [BRW_CONDITIONAL_L] = ".l", + [BRW_CONDITIONAL_LE] = ".le", + [BRW_CONDITIONAL_R] = ".r", + [BRW_CONDITIONAL_O] = ".o", + [BRW_CONDITIONAL_U] = ".u", +}; + +char *negate[2] = { + [0] = "", + [1] = "-", +}; + +char *_abs[2] = { + [0] = "", + [1] = "(abs)", +}; + +char *vert_stride[16] = { + [0] = "0", + [1] = "1", + [2] = "2", + [3] = "4", + [4] = "8", + [5] = "16", + [6] = "32", + [15] = "VxH", +}; + +char *width[8] = { + [0] = "1", + [1] = "2", + [2] = "4", + [3] = "8", + [4] = "16", +}; + +char *horiz_stride[4] = { + [0] = "0", + [1] = "1", + [2] = "2", + [3] = "4" +}; + +char *chan_sel[4] = { + [0] = "x", + [1] = "y", + [2] = "z", + [3] = "w", +}; + +char *dest_condmod[16] = { +}; + +char *debug_ctrl[2] = { + [0] = "", + [1] = ".breakpoint" +}; + +char *saturate[2] = { + [0] = "", + [1] = ".sat" +}; + +char *exec_size[8] = { + [0] = "1", + [1] = "2", + [2] = "4", + [3] = "8", + [4] = "16", + [5] = "32" +}; + +char *pred_inv[2] = { + [0] = "+", + [1] = "-" +}; + +char *pred_ctrl_align16[16] = { + [1] = "", + [2] = ".x", + [3] = ".y", + [4] = ".z", + [5] = ".w", + [6] = ".any4h", + [7] = ".all4h", +}; + +char *pred_ctrl_align1[16] = { + [1] = "", + [2] = ".anyv", + [3] = ".allv", + [4] = ".any2h", + [5] = ".all2h", + [6] = ".any4h", + [7] = ".all4h", + [8] = ".any8h", + [9] = ".all8h", + [10] = ".any16h", + [11] = ".all16h", +}; + +char *thread_ctrl[4] = { + [0] = "", + [2] = "switch" +}; + +char *compr_ctrl[4] = { + [0] = "", + [1] = "sechalf", + [2] = "compr", +}; + +char *dep_ctrl[4] = { + [0] = "", + [1] = "NoDDClr", + [2] = "NoDDChk", + [3] = "NoDDClr,NoDDChk", +}; + +char *mask_ctrl[4] = { + [0] = "", + [1] = "nomask", +}; + +char *access_mode[2] = { + [0] = "align1", + [1] = "align16", +}; + +char *reg_encoding[8] = { + [0] = "UD", + [1] = "D", + [2] = "UW", + [3] = "W", + [4] = "UB", + [5] = "B", + [7] = "F" +}; + +char *imm_encoding[8] = { + [0] = "UD", + [1] = "D", + [2] = "UW", + [3] = "W", + [5] = "VF", + [5] = "V", + [7] = "F" +}; + +char *reg_file[4] = { + [0] = "A", + [1] = "g", + [2] = "m", + [3] = "imm", +}; + +char *writemask[16] = { + [0x0] = ".", + [0x1] = ".x", + [0x2] = ".y", + [0x3] = ".xy", + [0x4] = ".z", + [0x5] = ".xz", + [0x6] = ".yz", + [0x7] = ".xyz", + [0x8] = ".w", + [0x9] = ".xw", + [0xa] = ".yw", + [0xb] = ".xyw", + [0xc] = ".zw", + [0xd] = ".xzw", + [0xe] = ".yzw", + [0xf] = "", +}; + +char *end_of_thread[2] = { + [0] = "", + [1] = "EOT" +}; + +char *target_function[16] = { + [BRW_MESSAGE_TARGET_NULL] = "null", + [BRW_MESSAGE_TARGET_MATH] = "math", + [BRW_MESSAGE_TARGET_SAMPLER] = "sampler", + [BRW_MESSAGE_TARGET_GATEWAY] = "gateway", + [BRW_MESSAGE_TARGET_DATAPORT_READ] = "read", + [BRW_MESSAGE_TARGET_DATAPORT_WRITE] = "write", + [BRW_MESSAGE_TARGET_URB] = "urb", + [BRW_MESSAGE_TARGET_THREAD_SPAWNER] = "thread_spawner" +}; + +char *math_function[16] = { + [BRW_MATH_FUNCTION_INV] = "inv", + [BRW_MATH_FUNCTION_LOG] = "log", + [BRW_MATH_FUNCTION_EXP] = "exp", + [BRW_MATH_FUNCTION_SQRT] = "sqrt", + [BRW_MATH_FUNCTION_RSQ] = "rsq", + [BRW_MATH_FUNCTION_SIN] = "sin", + [BRW_MATH_FUNCTION_COS] = "cos", + [BRW_MATH_FUNCTION_SINCOS] = "sincos", + [BRW_MATH_FUNCTION_TAN] = "tan", + [BRW_MATH_FUNCTION_POW] = "pow", + [BRW_MATH_FUNCTION_INT_DIV_QUOTIENT_AND_REMAINDER] = "intdivmod", + [BRW_MATH_FUNCTION_INT_DIV_QUOTIENT] = "intmod", + [BRW_MATH_FUNCTION_INT_DIV_REMAINDER] = "intdiv", +}; + +char *math_saturate[2] = { + [0] = "", + [1] = "sat" +}; + +char *math_signed[2] = { + [0] = "", + [1] = "signed" +}; + +char *math_scalar[2] = { + [0] = "", + [1] = "scalar" +}; + +char *math_precision[2] = { + [0] = "", + [1] = "partial_precision" +}; + +char *urb_swizzle[4] = { + [BRW_URB_SWIZZLE_NONE] = "", + [BRW_URB_SWIZZLE_INTERLEAVE] = "interleave", + [BRW_URB_SWIZZLE_TRANSPOSE] = "transpose", +}; + +char *urb_allocate[2] = { + [0] = "", + [1] = "allocate" +}; + +char *urb_used[2] = { + [0] = "", + [1] = "used" +}; + +char *urb_complete[2] = { + [0] = "", + [1] = "complete" +}; + +char *sampler_target_format[4] = { + [0] = "F", + [2] = "UD", + [3] = "D" +}; + + +static int column; + +static int string (FILE *file, char *string) +{ + fputs (string, file); + column += strlen (string); + return 0; +} + +static int format (FILE *f, char *format, ...) +{ + char buf[1024]; + va_list args; + va_start (args, format); + + vsnprintf (buf, sizeof (buf) - 1, format, args); + string (f, buf); + return 0; +} + +static int newline (FILE *f) +{ + putc ('\n', f); + column = 0; + return 0; +} + +static int pad (FILE *f, int c) +{ + do + string (f, " "); + while (column < c); + return 0; +} + +static int control (FILE *file, char *name, char *ctrl[], GLuint id, int *space) +{ + if (!ctrl[id]) { + fprintf (file, "*** invalid %s value %d ", + name, id); + return 1; + } + if (ctrl[id][0]) + { + if (space && *space) + string (file, " "); + string (file, ctrl[id]); + if (space) + *space = 1; + } + return 0; +} + +static int print_opcode (FILE *file, int id) +{ + if (!opcode[id].name) { + format (file, "*** invalid opcode value %d ", id); + return 1; + } + string (file, opcode[id].name); + return 0; +} + +static int reg (FILE *file, GLuint _reg_file, GLuint _reg_nr) +{ + int err = 0; + if (_reg_file == BRW_ARCHITECTURE_REGISTER_FILE) { + switch (_reg_nr & 0xf0) { + case BRW_ARF_NULL: + string (file, "null"); + return -1; + case BRW_ARF_ADDRESS: + format (file, "a%d", _reg_nr & 0x0f); + break; + case BRW_ARF_ACCUMULATOR: + format (file, "acc%d", _reg_nr & 0x0f); + break; + case BRW_ARF_MASK: + format (file, "mask%d", _reg_nr & 0x0f); + break; + case BRW_ARF_MASK_STACK: + format (file, "msd%d", _reg_nr & 0x0f); + break; + case BRW_ARF_STATE: + format (file, "sr%d", _reg_nr & 0x0f); + break; + case BRW_ARF_CONTROL: + format (file, "cr%d", _reg_nr & 0x0f); + break; + case BRW_ARF_NOTIFICATION_COUNT: + format (file, "n%d", _reg_nr & 0x0f); + break; + case BRW_ARF_IP: + string (file, "ip"); + return -1; + break; + default: + format (file, "ARF%d", _reg_nr); + break; + } + } else { + err |= control (file, "src reg file", reg_file, _reg_file, NULL); + format (file, "%d", _reg_nr); + } + return err; +} + +static int dest (FILE *file, struct brw_instruction *inst) +{ + int err = 0; + + if (inst->header.access_mode == BRW_ALIGN_1) + { + if (inst->bits1.da1.dest_address_mode == BRW_ADDRESS_DIRECT) + { + err |= reg (file, inst->bits1.da1.dest_reg_file, inst->bits1.da1.dest_reg_nr); + if (err == -1) + return 0; + if (inst->bits1.da1.dest_subreg_nr) + format (file, ".%d", inst->bits1.da1.dest_subreg_nr); + format (file, "<%d>", inst->bits1.da1.dest_horiz_stride); + err |= control (file, "dest reg encoding", reg_encoding, inst->bits1.da1.dest_reg_type, NULL); + } + else + { + string (file, "g[a0"); + if (inst->bits1.ia1.dest_subreg_nr) + format (file, ".%d", inst->bits1.ia1.dest_subreg_nr); + if (inst->bits1.ia1.dest_indirect_offset) + format (file, " %d", inst->bits1.ia1.dest_indirect_offset); + string (file, "]"); + format (file, "<%d>", inst->bits1.ia1.dest_horiz_stride); + err |= control (file, "dest reg encoding", reg_encoding, inst->bits1.ia1.dest_reg_type, NULL); + } + } + else + { + if (inst->bits1.da16.dest_address_mode == BRW_ADDRESS_DIRECT) + { + err |= reg (file, inst->bits1.da16.dest_reg_file, inst->bits1.da16.dest_reg_nr); + if (err == -1) + return 0; + if (inst->bits1.da16.dest_subreg_nr) + format (file, ".%d", inst->bits1.da16.dest_subreg_nr); + string (file, "<1>"); + err |= control (file, "writemask", writemask, inst->bits1.da16.dest_writemask, NULL); + err |= control (file, "dest reg encoding", reg_encoding, inst->bits1.da16.dest_reg_type, NULL); + } + else + { + err = 1; + string (file, "Indirect align16 address mode not supported"); + } + } + + return 0; +} + +static int src_align1_region (FILE *file, + GLuint _vert_stride, GLuint _width, GLuint _horiz_stride) +{ + int err = 0; + string (file, "<"); + err |= control (file, "vert stride", vert_stride, _vert_stride, NULL); + string (file, ","); + err |= control (file, "width", width, _width, NULL); + string (file, ","); + err |= control (file, "horiz_stride", horiz_stride, _horiz_stride, NULL); + string (file, ">"); + return err; +} + +static int src_da1 (FILE *file, GLuint type, GLuint _reg_file, + GLuint _vert_stride, GLuint _width, GLuint _horiz_stride, + GLuint reg_num, GLuint sub_reg_num, GLuint __abs, GLuint _negate) +{ + int err = 0; + err |= control (file, "negate", negate, _negate, NULL); + err |= control (file, "abs", _abs, __abs, NULL); + + err |= reg (file, _reg_file, reg_num); + if (err == -1) + return 0; + if (sub_reg_num) + format (file, ".%d", sub_reg_num); + src_align1_region (file, _vert_stride, _width, _horiz_stride); + err |= control (file, "src reg encoding", reg_encoding, type, NULL); + return err; +} + +static int src_ia1 (FILE *file, + GLuint type, + GLuint _reg_file, + GLint _addr_imm, + GLuint _addr_subreg_nr, + GLuint _negate, + GLuint __abs, + GLuint _addr_mode, + GLuint _horiz_stride, + GLuint _width, + GLuint _vert_stride) +{ + int err = 0; + err |= control (file, "negate", negate, _negate, NULL); + err |= control (file, "abs", _abs, __abs, NULL); + + string (file, "g[a0"); + if (_addr_subreg_nr) + format (file, ".%d", _addr_subreg_nr); + if (_addr_imm) + format (file, " %d", _addr_imm); + string (file, "]"); + src_align1_region (file, _vert_stride, _width, _horiz_stride); + err |= control (file, "src reg encoding", reg_encoding, type, NULL); + return err; +} + +static int src_da16 (FILE *file, + GLuint _reg_type, + GLuint _reg_file, + GLuint _vert_stride, + GLuint _reg_nr, + GLuint _subreg_nr, + GLuint __abs, + GLuint _negate, + GLuint swz_x, + GLuint swz_y, + GLuint swz_z, + GLuint swz_w) +{ + int err = 0; + err |= control (file, "negate", negate, _negate, NULL); + err |= control (file, "abs", _abs, __abs, NULL); + + err |= reg (file, _reg_file, _reg_nr); + if (err == -1) + return 0; + if (_subreg_nr) + format (file, ".%d", _subreg_nr); + string (file, "<"); + err |= control (file, "vert stride", vert_stride, _vert_stride, NULL); + string (file, ",1,1>"); + err |= control (file, "src da16 reg type", reg_encoding, _reg_type, NULL); + /* + * Three kinds of swizzle display: + * identity - nothing printed + * 1->all - print the single channel + * 1->1 - print the mapping + */ + if (swz_x == BRW_CHANNEL_X && + swz_y == BRW_CHANNEL_Y && + swz_z == BRW_CHANNEL_Z && + swz_w == BRW_CHANNEL_W) + { + ; + } + else if (swz_x == swz_y && swz_x == swz_z && swz_x == swz_w) + { + string (file, "."); + err |= control (file, "channel select", chan_sel, swz_x, NULL); + } + else + { + string (file, "."); + err |= control (file, "channel select", chan_sel, swz_x, NULL); + err |= control (file, "channel select", chan_sel, swz_y, NULL); + err |= control (file, "channel select", chan_sel, swz_z, NULL); + err |= control (file, "channel select", chan_sel, swz_w, NULL); + } + return err; +} + + +static int imm (FILE *file, GLuint type, struct brw_instruction *inst) { + switch (type) { + case BRW_REGISTER_TYPE_UD: + format (file, "0x%08xUD", inst->bits3.ud); + break; + case BRW_REGISTER_TYPE_D: + format (file, "%dD", inst->bits3.d); + break; + case BRW_REGISTER_TYPE_UW: + format (file, "0x%04xUW", (uint16_t) inst->bits3.ud); + break; + case BRW_REGISTER_TYPE_W: + format (file, "%dW", (int16_t) inst->bits3.d); + break; + case BRW_REGISTER_TYPE_UB: + format (file, "0x%02xUB", (int8_t) inst->bits3.ud); + break; + case BRW_REGISTER_TYPE_VF: + format (file, "Vector Float"); + break; + case BRW_REGISTER_TYPE_V: + format (file, "0x%08xV", inst->bits3.ud); + break; + case BRW_REGISTER_TYPE_F: + format (file, "%-gF", inst->bits3.f); + } + return 0; +} + +static int src0 (FILE *file, struct brw_instruction *inst) +{ + if (inst->bits1.da1.src0_reg_file == BRW_IMMEDIATE_VALUE) + return imm (file, inst->bits1.da1.src0_reg_type, + inst); + else if (inst->header.access_mode == BRW_ALIGN_1) + { + if (inst->bits2.da1.src0_address_mode == BRW_ADDRESS_DIRECT) + { + return src_da1 (file, + inst->bits1.da1.src0_reg_type, + inst->bits1.da1.src0_reg_file, + inst->bits2.da1.src0_vert_stride, + inst->bits2.da1.src0_width, + inst->bits2.da1.src0_horiz_stride, + inst->bits2.da1.src0_reg_nr, + inst->bits2.da1.src0_subreg_nr, + inst->bits2.da1.src0_abs, + inst->bits2.da1.src0_negate); + } + else + { + return src_ia1 (file, + inst->bits1.ia1.src0_reg_type, + inst->bits1.ia1.src0_reg_file, + inst->bits2.ia1.src0_indirect_offset, + inst->bits2.ia1.src0_subreg_nr, + inst->bits2.ia1.src0_negate, + inst->bits2.ia1.src0_abs, + inst->bits2.ia1.src0_address_mode, + inst->bits2.ia1.src0_horiz_stride, + inst->bits2.ia1.src0_width, + inst->bits2.ia1.src0_vert_stride); + } + } + else + { + if (inst->bits2.da16.src0_address_mode == BRW_ADDRESS_DIRECT) + { + return src_da16 (file, + inst->bits1.da16.src0_reg_type, + inst->bits1.da16.src0_reg_file, + inst->bits2.da16.src0_vert_stride, + inst->bits2.da16.src0_reg_nr, + inst->bits2.da16.src0_subreg_nr, + inst->bits2.da16.src0_abs, + inst->bits2.da16.src0_negate, + inst->bits2.da16.src0_swz_x, + inst->bits2.da16.src0_swz_y, + inst->bits2.da16.src0_swz_z, + inst->bits2.da16.src0_swz_w); + } + else + { + string (file, "Indirect align16 address mode not supported"); + return 1; + } + } +} + +static int src1 (FILE *file, struct brw_instruction *inst) +{ + if (inst->bits1.da1.src1_reg_file == BRW_IMMEDIATE_VALUE) + return imm (file, inst->bits1.da1.src1_reg_type, + inst); + else if (inst->header.access_mode == BRW_ALIGN_1) + { + if (inst->bits3.da1.src1_address_mode == BRW_ADDRESS_DIRECT) + { + return src_da1 (file, + inst->bits1.da1.src1_reg_type, + inst->bits1.da1.src1_reg_file, + inst->bits3.da1.src1_vert_stride, + inst->bits3.da1.src1_width, + inst->bits3.da1.src1_horiz_stride, + inst->bits3.da1.src1_reg_nr, + inst->bits3.da1.src1_subreg_nr, + inst->bits3.da1.src1_abs, + inst->bits3.da1.src1_negate); + } + else + { + return src_ia1 (file, + inst->bits1.ia1.src1_reg_type, + inst->bits1.ia1.src1_reg_file, + inst->bits3.ia1.src1_indirect_offset, + inst->bits3.ia1.src1_subreg_nr, + inst->bits3.ia1.src1_negate, + inst->bits3.ia1.src1_abs, + inst->bits3.ia1.src1_address_mode, + inst->bits3.ia1.src1_horiz_stride, + inst->bits3.ia1.src1_width, + inst->bits3.ia1.src1_vert_stride); + } + } + else + { + if (inst->bits3.da16.src1_address_mode == BRW_ADDRESS_DIRECT) + { + return src_da16 (file, + inst->bits1.da16.src1_reg_type, + inst->bits1.da16.src1_reg_file, + inst->bits3.da16.src1_vert_stride, + inst->bits3.da16.src1_reg_nr, + inst->bits3.da16.src1_subreg_nr, + inst->bits3.da16.src1_abs, + inst->bits3.da16.src1_negate, + inst->bits3.da16.src1_swz_x, + inst->bits3.da16.src1_swz_y, + inst->bits3.da16.src1_swz_z, + inst->bits3.da16.src1_swz_w); + } + else + { + string (file, "Indirect align16 address mode not supported"); + return 1; + } + } +} + +int brw_disasm (FILE *file, struct brw_instruction *inst) +{ + int err = 0; + int space = 0; + + if (inst->header.predicate_control) { + string (file, "("); + err |= control (file, "predicate inverse", pred_inv, inst->header.predicate_inverse, NULL); + string (file, "f0"); + if (inst->bits2.da1.flag_reg_nr) + format (file, ".%d", inst->bits2.da1.flag_reg_nr); + if (inst->header.access_mode == BRW_ALIGN_1) + err |= control (file, "predicate control align1", pred_ctrl_align1, + inst->header.predicate_control, NULL); + else + err |= control (file, "predicate control align16", pred_ctrl_align16, + inst->header.predicate_control, NULL); + string (file, ") "); + } + + err |= print_opcode (file, inst->header.opcode); + err |= control (file, "saturate", saturate, inst->header.saturate, NULL); + err |= control (file, "debug control", debug_ctrl, inst->header.debug_control, NULL); + + if (inst->header.opcode != BRW_OPCODE_SEND) + err |= control (file, "conditional modifier", conditional_modifier, + inst->header.destreg__conditionalmod, NULL); + + if (inst->header.opcode != BRW_OPCODE_NOP) { + string (file, "("); + err |= control (file, "execution size", exec_size, inst->header.execution_size, NULL); + string (file, ")"); + } + + if (inst->header.opcode == BRW_OPCODE_SEND) + format (file, " %d", inst->header.destreg__conditionalmod); + + if (opcode[inst->header.opcode].ndst > 0) { + pad (file, 16); + err |= dest (file, inst); + } + if (opcode[inst->header.opcode].nsrc > 0) { + pad (file, 32); + err |= src0 (file, inst); + } + if (opcode[inst->header.opcode].nsrc > 1) { + pad (file, 48); + err |= src1 (file, inst); + } + + if (inst->header.opcode == BRW_OPCODE_SEND) { + newline (file); + pad (file, 16); + space = 0; + err |= control (file, "target function", target_function, + inst->bits3.generic.msg_target, &space); + switch (inst->bits3.generic.msg_target) { + case BRW_MESSAGE_TARGET_MATH: + err |= control (file, "math function", math_function, + inst->bits3.math.function, &space); + err |= control (file, "math saturate", math_saturate, + inst->bits3.math.saturate, &space); + err |= control (file, "math signed", math_signed, + inst->bits3.math.int_type, &space); + err |= control (file, "math scalar", math_scalar, + inst->bits3.math.data_type, &space); + err |= control (file, "math precision", math_precision, + inst->bits3.math.precision, &space); + break; + case BRW_MESSAGE_TARGET_SAMPLER: + format (file, " (%d, %d, ", + inst->bits3.sampler.binding_table_index, + inst->bits3.sampler.sampler); + err |= control (file, "sampler target format", sampler_target_format, + inst->bits3.sampler.return_format, NULL); + string (file, ")"); + break; + case BRW_MESSAGE_TARGET_DATAPORT_WRITE: + format (file, " (%d, %d, %d, %d)", + inst->bits3.dp_write.binding_table_index, + (inst->bits3.dp_write.pixel_scoreboard_clear << 3) | + inst->bits3.dp_write.msg_control, + inst->bits3.dp_write.msg_type, + inst->bits3.dp_write.send_commit_msg); + break; + case BRW_MESSAGE_TARGET_URB: + format (file, " %d", inst->bits3.urb.offset); + space = 1; + err |= control (file, "urb swizzle", urb_swizzle, + inst->bits3.urb.swizzle_control, &space); + err |= control (file, "urb allocate", urb_allocate, + inst->bits3.urb.allocate, &space); + err |= control (file, "urb used", urb_used, + inst->bits3.urb.used, &space); + err |= control (file, "urb complete", urb_complete, + inst->bits3.urb.complete, &space); + break; + case BRW_MESSAGE_TARGET_THREAD_SPAWNER: + break; + default: + format (file, "unsupported target %d", inst->bits3.generic.msg_target); + break; + } + if (space) + string (file, " "); + format (file, "mlen %d", + inst->bits3.generic.msg_length); + format (file, " rlen %d", + inst->bits3.generic.response_length); + } + pad (file, 64); + if (inst->header.opcode != BRW_OPCODE_NOP) { + string (file, "{"); + space = 1; + err |= control(file, "access mode", access_mode, inst->header.access_mode, &space); + err |= control (file, "mask control", mask_ctrl, inst->header.mask_control, &space); + err |= control (file, "dependency control", dep_ctrl, inst->header.dependency_control, &space); + err |= control (file, "compression control", compr_ctrl, inst->header.compression_control, &space); + err |= control (file, "thread control", thread_ctrl, inst->header.thread_control, &space); + if (inst->header.opcode == BRW_OPCODE_SEND) + err |= control (file, "end of thread", end_of_thread, + inst->bits3.generic.end_of_thread, &space); + if (space) + string (file, " "); + string (file, "}"); + } + string (file, ";"); + newline (file); + return err; +} diff --git a/src/mesa/drivers/dri/i965/brw_draw.c b/src/mesa/drivers/dri/i965/brw_draw.c index 5152c3f3a5f..682094ff139 100644 --- a/src/mesa/drivers/dri/i965/brw_draw.c +++ b/src/mesa/drivers/dri/i965/brw_draw.c @@ -141,6 +141,8 @@ static void brw_emit_prim(struct brw_context *brw, prim_packet.verts_per_instance = trim(prim->mode, prim->count); prim_packet.start_vert_location = prim->start; + if (prim->indexed) + prim_packet.start_vert_location += brw->ib.start_vertex_offset; prim_packet.instance_count = 1; prim_packet.start_instance_location = 0; prim_packet.base_vert_location = 0; @@ -422,54 +424,31 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx, return retval; } -static GLboolean brw_need_rebase( GLcontext *ctx, - const struct gl_client_array *arrays[], - const struct _mesa_index_buffer *ib, - GLuint min_index ) -{ - if (min_index == 0) - return GL_FALSE; - - if (ib) { - if (!vbo_all_varyings_in_vbos(arrays)) - return GL_TRUE; - else - return GL_FALSE; - } - else { - /* Hmm. This isn't quite what I wanted. BRW can actually - * handle the mixed case well enough that we shouldn't need to - * rebase. However, it's probably not very common, nor hugely - * expensive to do it this way: - */ - if (!vbo_all_varyings_in_vbos(arrays)) - return GL_TRUE; - else - return GL_FALSE; - } -} - - void brw_draw_prims( GLcontext *ctx, const struct gl_client_array *arrays[], const struct _mesa_prim *prim, GLuint nr_prims, const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, GLuint min_index, GLuint max_index ) { GLboolean retval; - /* Decide if we want to rebase. If so we end up recursing once - * only into this function. - */ - if (brw_need_rebase( ctx, arrays, ib, min_index )) { - vbo_rebase_prims( ctx, arrays, - prim, nr_prims, - ib, min_index, max_index, - brw_draw_prims ); - - return; + if (!vbo_all_varyings_in_vbos(arrays)) { + if (!index_bounds_valid) + vbo_get_minmax_index(ctx, prim, ib, &min_index, &max_index); + + /* Decide if we want to rebase. If so we end up recursing once + * only into this function. + */ + if (min_index != 0) { + vbo_rebase_prims(ctx, arrays, + prim, nr_prims, + ib, min_index, max_index, + brw_draw_prims ); + return; + } } /* Make a first attempt at drawing: diff --git a/src/mesa/drivers/dri/i965/brw_draw.h b/src/mesa/drivers/dri/i965/brw_draw.h index 9aebbdb1b86..2a14db217fc 100644 --- a/src/mesa/drivers/dri/i965/brw_draw.h +++ b/src/mesa/drivers/dri/i965/brw_draw.h @@ -39,6 +39,7 @@ void brw_draw_prims( GLcontext *ctx, const struct _mesa_prim *prims, GLuint nr_prims, const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, GLuint min_index, GLuint max_index ); diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c index c29f1dd5c03..ab6b62812f1 100644 --- a/src/mesa/drivers/dri/i965/brw_draw_upload.c +++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c @@ -350,9 +350,6 @@ static void brw_prepare_vertices(struct brw_context *brw) unsigned int min_index = brw->vb.min_index; unsigned int max_index = brw->vb.max_index; - struct brw_vertex_element *enabled[VERT_ATTRIB_MAX]; - GLuint nr_enabled = 0; - struct brw_vertex_element *upload[VERT_ATTRIB_MAX]; GLuint nr_uploads = 0; @@ -362,12 +359,13 @@ static void brw_prepare_vertices(struct brw_context *brw) _mesa_printf("%s %d..%d\n", __FUNCTION__, min_index, max_index); /* Accumulate the list of enabled arrays. */ + brw->vb.nr_enabled = 0; while (vs_inputs) { GLuint i = _mesa_ffsll(vs_inputs) - 1; struct brw_vertex_element *input = &brw->vb.inputs[i]; vs_inputs &= ~(1 << i); - enabled[nr_enabled++] = input; + brw->vb.enabled[brw->vb.nr_enabled++] = input; } /* XXX: In the rare cases where this happens we fallback all @@ -376,16 +374,15 @@ static void brw_prepare_vertices(struct brw_context *brw) * cases with > 17 vertex attributes enabled, so it probably * isn't an issue at this point. */ - if (nr_enabled >= BRW_VEP_MAX) { + if (brw->vb.nr_enabled >= BRW_VEP_MAX) { intel->Fallback = 1; return; } - for (i = 0; i < nr_enabled; i++) { - struct brw_vertex_element *input = enabled[i]; + for (i = 0; i < brw->vb.nr_enabled; i++) { + struct brw_vertex_element *input = brw->vb.enabled[i]; input->element_size = get_size(input->glarray->Type) * input->glarray->Size; - input->count = input->glarray->StrideB ? max_index + 1 - min_index : 1; if (input->glarray->BufferObj->Name != 0) { struct intel_buffer_object *intel_buffer = @@ -398,7 +395,23 @@ static void brw_prepare_vertices(struct brw_context *brw) dri_bo_reference(input->bo); input->offset = (unsigned long)input->glarray->Ptr; input->stride = input->glarray->StrideB; + input->count = input->glarray->_MaxElement; + + /* This is a common place to reach if the user mistakenly supplies + * a pointer in place of a VBO offset. If we just let it go through, + * we may end up dereferencing a pointer beyond the bounds of the + * GTT. We would hope that the VBO's max_index would save us, but + * Mesa appears to hand us min/max values not clipped to the + * array object's _MaxElement, and _MaxElement frequently appears + * to be wrong anyway. + * + * The VBO spec allows application termination in this case, and it's + * probably a service to the poor programmer to do so rather than + * trying to just not render. + */ + assert(input->offset < input->bo->size); } else { + input->count = input->glarray->StrideB ? max_index + 1 - min_index : 1; if (input->bo != NULL) { /* Already-uploaded vertex data is present from a previous * prepare_vertices, but we had to re-validate state due to @@ -466,8 +479,8 @@ static void brw_prepare_vertices(struct brw_context *brw) brw_prepare_query_begin(brw); - for (i = 0; i < nr_enabled; i++) { - struct brw_vertex_element *input = enabled[i]; + for (i = 0; i < brw->vb.nr_enabled; i++) { + struct brw_vertex_element *input = brw->vb.enabled[i]; brw_add_validated_bo(brw, input->bo); } @@ -477,34 +490,44 @@ static void brw_emit_vertices(struct brw_context *brw) { GLcontext *ctx = &brw->intel.ctx; struct intel_context *intel = intel_context(ctx); - GLbitfield vs_inputs = brw->vs.prog_data->inputs_read; - struct brw_vertex_element *enabled[VERT_ATTRIB_MAX]; GLuint i; - GLuint nr_enabled = 0; - /* Accumulate the list of enabled arrays. */ - while (vs_inputs) { - i = _mesa_ffsll(vs_inputs) - 1; - struct brw_vertex_element *input = &brw->vb.inputs[i]; + brw_emit_query_begin(brw); - vs_inputs &= ~(1 << i); - enabled[nr_enabled++] = input; + /* If the VS doesn't read any inputs (calculating vertex position from + * a state variable for some reason, for example), emit a single pad + * VERTEX_ELEMENT struct and bail. + * + * The stale VB state stays in place, but they don't do anything unless + * a VE loads from them. + */ + if (brw->vb.nr_enabled == 0) { + BEGIN_BATCH(3, IGNORE_CLIPRECTS); + OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | 1); + OUT_BATCH((0 << BRW_VE0_INDEX_SHIFT) | + BRW_VE0_VALID | + (BRW_SURFACEFORMAT_R32G32B32A32_FLOAT << BRW_VE0_FORMAT_SHIFT) | + (0 << BRW_VE0_SRC_OFFSET_SHIFT)); + OUT_BATCH((BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_0_SHIFT) | + (BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_1_SHIFT) | + (BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_2_SHIFT) | + (BRW_VE1_COMPONENT_STORE_1_FLT << BRW_VE1_COMPONENT_3_SHIFT)); + ADVANCE_BATCH(); + return; } - brw_emit_query_begin(brw); - /* Now emit VB and VEP state packets. * * This still defines a hardware VB for each input, even if they * are interleaved or from the same VBO. TBD if this makes a * performance difference. */ - BEGIN_BATCH(1 + nr_enabled * 4, IGNORE_CLIPRECTS); + BEGIN_BATCH(1 + brw->vb.nr_enabled * 4, IGNORE_CLIPRECTS); OUT_BATCH((CMD_VERTEX_BUFFER << 16) | - ((1 + nr_enabled * 4) - 2)); + ((1 + brw->vb.nr_enabled * 4) - 2)); - for (i = 0; i < nr_enabled; i++) { - struct brw_vertex_element *input = enabled[i]; + for (i = 0; i < brw->vb.nr_enabled; i++) { + struct brw_vertex_element *input = brw->vb.enabled[i]; OUT_BATCH((i << BRW_VB0_INDEX_SHIFT) | BRW_VB0_ACCESS_VERTEXDATA | @@ -524,15 +547,15 @@ static void brw_emit_vertices(struct brw_context *brw) input->offset + input->element_size); } } else - OUT_BATCH(brw->vb.max_index); + OUT_BATCH(input->count); OUT_BATCH(0); /* Instance data step rate */ } ADVANCE_BATCH(); - BEGIN_BATCH(1 + nr_enabled * 2, IGNORE_CLIPRECTS); - OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | ((1 + nr_enabled * 2) - 2)); - for (i = 0; i < nr_enabled; i++) { - struct brw_vertex_element *input = enabled[i]; + BEGIN_BATCH(1 + brw->vb.nr_enabled * 2, IGNORE_CLIPRECTS); + OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | ((1 + brw->vb.nr_enabled * 2) - 2)); + for (i = 0; i < brw->vb.nr_enabled; i++) { + struct brw_vertex_element *input = brw->vb.enabled[i]; uint32_t format = get_surface_type(input->glarray->Type, input->glarray->Size, input->glarray->Format, @@ -589,17 +612,20 @@ static void brw_prepare_indices(struct brw_context *brw) dri_bo *bo = NULL; struct gl_buffer_object *bufferobj; GLuint offset; + GLuint ib_type_size; if (index_buffer == NULL) return; - ib_size = get_size(index_buffer->type) * index_buffer->count; + ib_type_size = get_size(index_buffer->type); + ib_size = ib_type_size * index_buffer->count; bufferobj = index_buffer->obj;; /* Turn into a proper VBO: */ if (!bufferobj->Name) { - + brw->ib.start_vertex_offset = 0; + /* Get new bufferobj, offset: */ get_space(brw, ib_size, &bo, &offset); @@ -615,6 +641,7 @@ static void brw_prepare_indices(struct brw_context *brw) } } else { offset = (GLuint) (unsigned long) index_buffer->ptr; + brw->ib.start_vertex_offset = 0; /* If the index buffer isn't aligned to its element size, we have to * rebase it into a temporary. @@ -635,39 +662,62 @@ static void brw_prepare_indices(struct brw_context *brw) bo = intel_bufferobj_buffer(intel, intel_buffer_object(bufferobj), INTEL_READ); dri_bo_reference(bo); + + /* Use CMD_3D_PRIM's start_vertex_offset to avoid re-uploading + * the index buffer state when we're just moving the start index + * of our drawing. + */ + brw->ib.start_vertex_offset = offset / ib_type_size; + offset = 0; + ib_size = bo->size; } } - dri_bo_unreference(brw->ib.bo); - brw->ib.bo = bo; - brw->ib.offset = offset; + if (brw->ib.bo != bo || + brw->ib.offset != offset || + brw->ib.size != ib_size) + { + drm_intel_bo_unreference(brw->ib.bo); + brw->ib.bo = bo; + brw->ib.offset = offset; + brw->ib.size = ib_size; + + brw->state.dirty.brw |= BRW_NEW_INDEX_BUFFER; + } else { + drm_intel_bo_unreference(bo); + } brw_add_validated_bo(brw, brw->ib.bo); } -static void brw_emit_indices(struct brw_context *brw) +const struct brw_tracked_state brw_indices = { + .dirty = { + .mesa = 0, + .brw = BRW_NEW_INDICES, + .cache = 0, + }, + .prepare = brw_prepare_indices, +}; + +static void brw_emit_index_buffer(struct brw_context *brw) { struct intel_context *intel = &brw->intel; const struct _mesa_index_buffer *index_buffer = brw->ib.ib; - GLuint ib_size; if (index_buffer == NULL) return; - ib_size = get_size(index_buffer->type) * index_buffer->count - 1; - /* Emit the indexbuffer packet: */ { struct brw_indexbuffer ib; memset(&ib, 0, sizeof(ib)); - + ib.header.bits.opcode = CMD_INDEX_BUFFER; ib.header.bits.length = sizeof(ib)/4 - 2; ib.header.bits.index_format = get_index_type(index_buffer->type); ib.header.bits.cut_index_enable = 0; - BEGIN_BATCH(4, IGNORE_CLIPRECTS); OUT_BATCH( ib.header.dword ); @@ -676,18 +726,17 @@ static void brw_emit_indices(struct brw_context *brw) brw->ib.offset); OUT_RELOC(brw->ib.bo, I915_GEM_DOMAIN_VERTEX, 0, - brw->ib.offset + ib_size); + brw->ib.offset + brw->ib.size); OUT_BATCH( 0 ); ADVANCE_BATCH(); } } -const struct brw_tracked_state brw_indices = { +const struct brw_tracked_state brw_index_buffer = { .dirty = { .mesa = 0, - .brw = BRW_NEW_BATCH | BRW_NEW_INDICES, + .brw = BRW_NEW_BATCH | BRW_NEW_INDEX_BUFFER, .cache = 0, }, - .prepare = brw_prepare_indices, - .emit = brw_emit_indices, + .emit = brw_emit_index_buffer, }; diff --git a/src/mesa/drivers/dri/i965/brw_eu.c b/src/mesa/drivers/dri/i965/brw_eu.c index c53efba5991..1df561386e6 100644 --- a/src/mesa/drivers/dri/i965/brw_eu.c +++ b/src/mesa/drivers/dri/i965/brw_eu.c @@ -62,7 +62,7 @@ void brw_set_predicate_control( struct brw_compile *p, GLuint pc ) void brw_set_conditionalmod( struct brw_compile *p, GLuint conditional ) { - p->current->header.destreg__conditonalmod = conditional; + p->current->header.destreg__conditionalmod = conditional; } void brw_set_access_mode( struct brw_compile *p, GLuint access_mode ) diff --git a/src/mesa/drivers/dri/i965/brw_eu_emit.c b/src/mesa/drivers/dri/i965/brw_eu_emit.c index 2412014248c..241cdc33f86 100644 --- a/src/mesa/drivers/dri/i965/brw_eu_emit.c +++ b/src/mesa/drivers/dri/i965/brw_eu_emit.c @@ -481,8 +481,8 @@ static struct brw_instruction *next_insn( struct brw_compile *p, /* Reset this one-shot flag: */ - if (p->current->header.destreg__conditonalmod) { - p->current->header.destreg__conditonalmod = 0; + if (p->current->header.destreg__conditionalmod) { + p->current->header.destreg__conditionalmod = 0; p->current->header.predicate_control = BRW_PREDICATE_NORMAL; } @@ -679,7 +679,7 @@ struct brw_instruction *brw_ELSE(struct brw_compile *p, assert(if_insn->header.opcode == BRW_OPCODE_IF); if_insn->bits3.if_else.jump_count = br * (insn - if_insn); - if_insn->bits3.if_else.pop_count = 1; + if_insn->bits3.if_else.pop_count = 0; if_insn->bits3.if_else.pad0 = 0; } @@ -871,7 +871,7 @@ void brw_CMP(struct brw_compile *p, { struct brw_instruction *insn = next_insn(p, BRW_OPCODE_CMP); - insn->header.destreg__conditonalmod = conditional; + insn->header.destreg__conditionalmod = conditional; brw_set_dest(insn, dest); brw_set_src0(insn, src0); brw_set_src1(insn, src1); @@ -915,7 +915,7 @@ void brw_math( struct brw_compile *p, * instructions. */ insn->header.predicate_control = 0; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_dest(insn, dest); brw_set_src0(insn, src); @@ -952,7 +952,7 @@ void brw_math_16( struct brw_compile *p, brw_set_compression_control(p, BRW_COMPRESSION_NONE); insn = next_insn(p, BRW_OPCODE_SEND); - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_dest(insn, dest); brw_set_src0(insn, src); @@ -969,7 +969,7 @@ void brw_math_16( struct brw_compile *p, */ insn = next_insn(p, BRW_OPCODE_SEND); insn->header.compression_control = BRW_COMPRESSION_2NDHALF; - insn->header.destreg__conditonalmod = msg_reg_nr+1; + insn->header.destreg__conditionalmod = msg_reg_nr+1; brw_set_dest(insn, offset(dest,1)); brw_set_src0(insn, src); @@ -1016,7 +1016,7 @@ void brw_dp_WRITE_16( struct brw_compile *p, insn->header.predicate_control = 0; /* XXX */ insn->header.compression_control = BRW_COMPRESSION_NONE; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_dest(insn, dest); brw_set_src0(insn, src); @@ -1062,7 +1062,7 @@ void brw_dp_READ_16( struct brw_compile *p, insn->header.predicate_control = 0; /* XXX */ insn->header.compression_control = BRW_COMPRESSION_NONE; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_dest(insn, dest); /* UW? */ brw_set_src0(insn, retype(brw_vec8_grf(0, 0), BRW_REGISTER_TYPE_UW)); @@ -1116,7 +1116,7 @@ void brw_dp_READ_4( struct brw_compile *p, insn->header.predicate_control = BRW_PREDICATE_NONE; insn->header.compression_control = BRW_COMPRESSION_NONE; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; insn->header.mask_control = BRW_MASK_DISABLE; /* cast dest to a uword[8] vector */ @@ -1190,7 +1190,7 @@ void brw_dp_READ_4_vs(struct brw_compile *p, insn->header.predicate_control = BRW_PREDICATE_NONE; insn->header.compression_control = BRW_COMPRESSION_NONE; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; insn->header.mask_control = BRW_MASK_DISABLE; /*insn->header.access_mode = BRW_ALIGN_16;*/ @@ -1224,7 +1224,7 @@ void brw_fb_WRITE(struct brw_compile *p, insn->header.predicate_control = 0; /* XXX */ insn->header.compression_control = BRW_COMPRESSION_NONE; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_dest(insn, dest); brw_set_src0(insn, src0); @@ -1322,7 +1322,7 @@ void brw_SAMPLE(struct brw_compile *p, insn->header.predicate_control = 0; /* XXX */ insn->header.compression_control = BRW_COMPRESSION_NONE; - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_dest(insn, dest); brw_set_src0(insn, src0); @@ -1375,7 +1375,7 @@ void brw_urb_WRITE(struct brw_compile *p, brw_set_src0(insn, src0); brw_set_src1(insn, brw_imm_d(0)); - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_urb_message(p->brw, insn, @@ -1410,7 +1410,7 @@ void brw_ff_sync(struct brw_compile *p, brw_set_src0(insn, src0); brw_set_src1(insn, brw_imm_d(0)); - insn->header.destreg__conditonalmod = msg_reg_nr; + insn->header.destreg__conditionalmod = msg_reg_nr; brw_set_ff_sync_message(p->brw, insn, diff --git a/src/mesa/drivers/dri/i965/brw_gs_emit.c b/src/mesa/drivers/dri/i965/brw_gs_emit.c index 980eac7646d..a9b2aa2eace 100644 --- a/src/mesa/drivers/dri/i965/brw_gs_emit.c +++ b/src/mesa/drivers/dri/i965/brw_gs_emit.c @@ -101,7 +101,7 @@ static void brw_gs_emit_vue(struct brw_gs_compile *c, BRW_URB_SWIZZLE_NONE); } -void brw_gs_ff_sync(struct brw_gs_compile *c, int num_prim) +static void brw_gs_ff_sync(struct brw_gs_compile *c, int num_prim) { struct brw_compile *p = &c->func; brw_MOV(p, get_element_ud(c->reg.R0, 1), brw_imm_ud(num_prim)); diff --git a/src/mesa/drivers/dri/i965/brw_sf_state.c b/src/mesa/drivers/dri/i965/brw_sf_state.c index b5f6371c82c..bc0f0760738 100644 --- a/src/mesa/drivers/dri/i965/brw_sf_state.c +++ b/src/mesa/drivers/dri/i965/brw_sf_state.c @@ -113,7 +113,7 @@ struct brw_sf_unit_key { unsigned int nr_urb_entries, urb_size, sfsize; - GLenum front_face, cull_face; + GLenum front_face, cull_face, provoking_vertex; unsigned scissor:1; unsigned line_smooth:1; unsigned point_sprite:1; @@ -153,6 +153,9 @@ sf_unit_populate_key(struct brw_context *brw, struct brw_sf_unit_key *key) key->point_size = CLAMP(ctx->Point.Size, ctx->Point.MinSize, ctx->Point.MaxSize); key->point_attenuated = ctx->Point._Attenuated; + /* _NEW_LIGHT */ + key->provoking_vertex = ctx->Light.ProvokingVertex; + key->render_to_fbo = brw->intel.ctx.DrawBuffer->Name != 0; } @@ -284,9 +287,15 @@ sf_unit_create_from_key(struct brw_context *brw, struct brw_sf_unit_key *key, /* might be BRW_NEW_PRIMITIVE if we have to adjust pv for polygons: */ - sf.sf7.trifan_pv = 2; - sf.sf7.linestrip_pv = 1; - sf.sf7.tristrip_pv = 2; + if (key->provoking_vertex == GL_LAST_VERTEX_CONVENTION) { + sf.sf7.trifan_pv = 2; + sf.sf7.linestrip_pv = 1; + sf.sf7.tristrip_pv = 2; + } else { + sf.sf7.trifan_pv = 1; + sf.sf7.linestrip_pv = 0; + sf.sf7.tristrip_pv = 0; + } sf.sf7.line_last_pixel_enable = 0; /* Set bias for OpenGL rasterization rules: @@ -300,6 +309,9 @@ sf_unit_create_from_key(struct brw_context *brw, struct brw_sf_unit_key *key, &sf, sizeof(sf), NULL, NULL); + /* STATE_PREFETCH command description describes this state as being + * something loaded through the GPE (L2 ISC), so it's INSTRUCTION domain. + */ /* Emit SF program relocation */ dri_bo_emit_reloc(bo, I915_GEM_DOMAIN_INSTRUCTION, 0, @@ -340,6 +352,7 @@ static void upload_sf_unit( struct brw_context *brw ) const struct brw_tracked_state brw_sf_unit = { .dirty = { .mesa = (_NEW_POLYGON | + _NEW_LIGHT | _NEW_LINE | _NEW_POINT | _NEW_SCISSOR | diff --git a/src/mesa/drivers/dri/i965/brw_state.h b/src/mesa/drivers/dri/i965/brw_state.h index bf9f6cae55e..78572356a3d 100644 --- a/src/mesa/drivers/dri/i965/brw_state.h +++ b/src/mesa/drivers/dri/i965/brw_state.h @@ -92,6 +92,7 @@ const struct brw_tracked_state brw_clear_batch_cache; const struct brw_tracked_state brw_drawing_rect; const struct brw_tracked_state brw_indices; const struct brw_tracked_state brw_vertices; +const struct brw_tracked_state brw_index_buffer; /** * Use same key for WM and VS surfaces. diff --git a/src/mesa/drivers/dri/i965/brw_state_upload.c b/src/mesa/drivers/dri/i965/brw_state_upload.c index 38d9dd8991e..95d42d2dcc5 100644 --- a/src/mesa/drivers/dri/i965/brw_state_upload.c +++ b/src/mesa/drivers/dri/i965/brw_state_upload.c @@ -94,6 +94,7 @@ const struct brw_tracked_state *atoms[] = &brw_drawing_rect, &brw_indices, + &brw_index_buffer, &brw_vertices, &brw_constant_buffer @@ -208,6 +209,7 @@ static struct dirty_bit_map brw_bits[] = { DEFINE_BIT(BRW_NEW_PSP), DEFINE_BIT(BRW_NEW_FENCE), DEFINE_BIT(BRW_NEW_INDICES), + DEFINE_BIT(BRW_NEW_INDEX_BUFFER), DEFINE_BIT(BRW_NEW_VERTICES), DEFINE_BIT(BRW_NEW_BATCH), DEFINE_BIT(BRW_NEW_DEPTH_BUFFER), diff --git a/src/mesa/drivers/dri/i965/brw_structs.h b/src/mesa/drivers/dri/i965/brw_structs.h index 8ba7eb27b36..a6de09207b7 100644 --- a/src/mesa/drivers/dri/i965/brw_structs.h +++ b/src/mesa/drivers/dri/i965/brw_structs.h @@ -1200,7 +1200,7 @@ struct brw_instruction GLuint predicate_control:4; GLuint predicate_inverse:1; GLuint execution_size:3; - GLuint destreg__conditonalmod:4; /* destreg - send, conditionalmod - others */ + GLuint destreg__conditionalmod:4; /* destreg - send, conditionalmod - others */ GLuint pad0:2; GLuint debug_control:1; GLuint saturate:1; @@ -1228,7 +1228,9 @@ struct brw_instruction GLuint dest_reg_type:3; GLuint src0_reg_file:2; GLuint src0_reg_type:3; - GLuint pad:6; + GLuint src1_reg_file:2; /* 0x00000c00 */ + GLuint src1_reg_type:3; /* 0x00007000 */ + GLuint pad:1; GLint dest_indirect_offset:10; /* offset against the deref'd address reg */ GLuint dest_subreg_nr:3; /* subnr for the address reg a0.x */ GLuint dest_horiz_stride:2; @@ -1243,7 +1245,7 @@ struct brw_instruction GLuint src0_reg_type:3; GLuint src1_reg_file:2; GLuint src1_reg_type:3; - GLuint pad0:1; + GLuint pad:1; GLuint dest_writemask:4; GLuint dest_subreg_nr:1; GLuint dest_reg_nr:8; @@ -1348,7 +1350,7 @@ struct brw_instruction GLuint src1_reg_nr:8; GLuint src1_abs:1; GLuint src1_negate:1; - GLuint pad:1; + GLuint src1_address_mode:1; GLuint src1_horiz_stride:2; GLuint src1_width:3; GLuint src1_vert_stride:4; @@ -1363,7 +1365,7 @@ struct brw_instruction GLuint src1_reg_nr:8; GLuint src1_abs:1; GLuint src1_negate:1; - GLuint pad0:1; + GLuint src1_address_mode:1; GLuint src1_swz_z:2; GLuint src1_swz_w:2; GLuint pad1:1; @@ -1377,7 +1379,7 @@ struct brw_instruction GLuint src1_subreg_nr:3; GLuint src1_abs:1; GLuint src1_negate:1; - GLuint pad0:1; + GLuint src1_address_mode:1; GLuint src1_horiz_stride:2; GLuint src1_width:3; GLuint src1_vert_stride:4; @@ -1565,6 +1567,7 @@ struct brw_instruction GLint d; GLuint ud; + float f; } bits3; }; diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c index 514f15d5e3a..83167b9258e 100644 --- a/src/mesa/drivers/dri/i965/brw_vs_emit.c +++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c @@ -68,6 +68,7 @@ static void release_tmps( struct brw_vs_compile *c ) static void brw_vs_alloc_regs( struct brw_vs_compile *c ) { GLuint i, reg = 0, mrf; + int attributes_in_vue; /* Determine whether to use a real constant buffer or use a block * of GRF registers for constants. The later is faster but only @@ -128,6 +129,11 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c ) reg++; } } + /* If there are no inputs, we'll still be reading one attribute's worth + * because it's required -- see urb_read_length setting. + */ + if (c->nr_inputs == 0) + reg++; /* Allocate outputs. The non-position outputs go straight into message regs. */ @@ -220,11 +226,22 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c ) * vertex urb, so is half the amount: */ c->prog_data.urb_read_length = (c->nr_inputs + 1) / 2; + /* Setting this field to 0 leads to undefined behavior according to the + * the VS_STATE docs. Our VUEs will always have at least one attribute + * sitting in them, even if it's padding. + */ + if (c->prog_data.urb_read_length == 0) + c->prog_data.urb_read_length = 1; + + /* The VS VUEs are shared by VF (outputting our inputs) and VS, so size + * them to fit the biggest thing they need to. + */ + attributes_in_vue = MAX2(c->nr_outputs, c->nr_inputs); if (BRW_IS_IGDNG(c->func.brw)) - c->prog_data.urb_entry_size = (c->nr_outputs + 6 + 3) / 4; + c->prog_data.urb_entry_size = (attributes_in_vue + 6 + 3) / 4; else - c->prog_data.urb_entry_size = (c->nr_outputs + 2 + 3) / 4; + c->prog_data.urb_entry_size = (attributes_in_vue + 2 + 3) / 4; c->prog_data.total_grf = reg; @@ -1245,9 +1262,30 @@ post_vs_emit( struct brw_vs_compile *c, /* patch up the END code to jump past subroutines, etc */ offset = last_inst - end_inst; - brw_set_src1(end_inst, brw_imm_d(offset * 16)); + if (offset > 1) { + brw_set_src1(end_inst, brw_imm_d(offset * 16)); + } else { + end_inst->header.opcode = BRW_OPCODE_NOP; + } } +static uint32_t +get_predicate(uint32_t swizzle) +{ + switch (swizzle) { + case SWIZZLE_XXXX: + return BRW_PREDICATE_ALIGN16_REPLICATE_X; + case SWIZZLE_YYYY: + return BRW_PREDICATE_ALIGN16_REPLICATE_Y; + case SWIZZLE_ZZZZ: + return BRW_PREDICATE_ALIGN16_REPLICATE_Z; + case SWIZZLE_WWWW: + return BRW_PREDICATE_ALIGN16_REPLICATE_W; + default: + _mesa_problem(NULL, "Unexpected predicate: 0x%08x\n", swizzle); + return BRW_PREDICATE_NORMAL; + } +} /* Emit the vertex program instructions here. */ @@ -1266,7 +1304,7 @@ void brw_vs_emit(struct brw_vs_compile *c ) GLuint file; if (INTEL_DEBUG & DEBUG_VS) { - _mesa_printf("vs-emit:\n"); + _mesa_printf("vs-mesa:\n"); _mesa_print_program(&c->vp->program.Base); _mesa_printf("\n"); } @@ -1453,7 +1491,10 @@ void brw_vs_emit(struct brw_vs_compile *c ) break; case OPCODE_IF: assert(if_depth < MAX_IF_DEPTH); - if_inst[if_depth++] = brw_IF(p, BRW_EXECUTE_8); + if_inst[if_depth] = brw_IF(p, BRW_EXECUTE_8); + if_inst[if_depth]->header.predicate_control = + get_predicate(inst->DstReg.CondSwizzle); + if_depth++; break; case OPCODE_ELSE: if_inst[if_depth-1] = brw_ELSE(p, if_inst[if_depth-1]); @@ -1541,6 +1582,19 @@ void brw_vs_emit(struct brw_vs_compile *c ) "unknown"); } + /* Set the predication update on the last instruction of the native + * instruction sequence. + * + * This would be problematic if it was set on a math instruction, + * but that shouldn't be the case with the current GLSL compiler. + */ + if (inst->CondUpdate) { + struct brw_instruction *hw_insn = &p->store[p->nr_insn - 1]; + + assert(hw_insn->header.destreg__conditionalmod == 0); + hw_insn->header.destreg__conditionalmod = BRW_CONDITIONAL_NZ; + } + if ((inst->DstReg.File == PROGRAM_OUTPUT) && (inst->DstReg.Index != VERT_RESULT_HPOS) && c->output_regs[inst->DstReg.Index].used_in_src) { @@ -1578,4 +1632,13 @@ void brw_vs_emit(struct brw_vs_compile *c ) emit_vertex_write(c); post_vs_emit(c, end_inst, last_inst); + + if (INTEL_DEBUG & DEBUG_VS) { + int i; + + _mesa_printf("vs-native:\n"); + for (i = 0; i < p->nr_insn; i++) + brw_disasm(stderr, &p->store[i]); + _mesa_printf("\n"); + } } diff --git a/src/mesa/drivers/dri/i965/brw_wm.c b/src/mesa/drivers/dri/i965/brw_wm.c index 14e05be4f6c..2292de94c44 100644 --- a/src/mesa/drivers/dri/i965/brw_wm.c +++ b/src/mesa/drivers/dri/i965/brw_wm.c @@ -41,13 +41,13 @@ GLuint brw_wm_nr_args( GLuint opcode ) { switch (opcode) { case WM_FRONTFACING: - return 0; case WM_PIXELXY: + return 0; case WM_CINTERP: case WM_WPOSXY: + case WM_DELTAXY: return 1; case WM_LINTERP: - case WM_DELTAXY: case WM_PIXELW: return 2; case WM_FB_WRITE: @@ -171,9 +171,11 @@ static void do_wm_prog( struct brw_context *brw, * differently from "simple" shaders. */ if (fp->isGLSL) { + c->dispatch_width = 8; brw_wm_glsl_emit(brw, c); } else { + c->dispatch_width = 16; brw_wm_non_glsl_emit(brw, c); } @@ -202,6 +204,7 @@ static void brw_wm_populate_key( struct brw_context *brw, /* BRW_NEW_FRAGMENT_PROGRAM */ const struct brw_fragment_program *fp = (struct brw_fragment_program *)brw->fragment_program; + GLboolean uses_depth = (fp->program.Base.InputsRead & (1 << FRAG_ATTRIB_WPOS)) != 0; GLuint lookup = 0; GLuint line_aa; GLuint i; @@ -263,6 +266,7 @@ static void brw_wm_populate_key( struct brw_context *brw, brw_wm_lookup_iz(line_aa, lookup, + uses_depth, key); diff --git a/src/mesa/drivers/dri/i965/brw_wm.h b/src/mesa/drivers/dri/i965/brw_wm.h index ba497432c60..ae98b5492db 100644 --- a/src/mesa/drivers/dri/i965/brw_wm.h +++ b/src/mesa/drivers/dri/i965/brw_wm.h @@ -260,6 +260,7 @@ struct brw_wm_compile { GLuint tmp_index; GLuint tmp_max; GLuint subroutines[BRW_WM_MAX_SUBROUTINE]; + GLuint dispatch_width; /** we may need up to 3 constants per instruction (if use_const_buffer) */ struct { @@ -292,6 +293,7 @@ void brw_wm_print_program( struct brw_wm_compile *c, void brw_wm_lookup_iz( GLuint line_aa, GLuint lookup, + GLboolean ps_uses_depth, struct brw_wm_prog_key *key ); GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp); diff --git a/src/mesa/drivers/dri/i965/brw_wm_emit.c b/src/mesa/drivers/dri/i965/brw_wm_emit.c index 9f82916c025..b3cf524c63e 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_emit.c +++ b/src/mesa/drivers/dri/i965/brw_wm_emit.c @@ -65,8 +65,7 @@ static INLINE struct brw_reg sechalf( struct brw_reg reg ) static void emit_pixel_xy(struct brw_compile *p, const struct brw_reg *dst, - GLuint mask, - const struct brw_reg *arg0) + GLuint mask) { struct brw_reg r1 = brw_vec1_grf(1, 0); struct brw_reg r1_uw = retype(r1, BRW_REGISTER_TYPE_UW); @@ -98,8 +97,7 @@ static void emit_pixel_xy(struct brw_compile *p, static void emit_delta_xy(struct brw_compile *p, const struct brw_reg *dst, GLuint mask, - const struct brw_reg *arg0, - const struct brw_reg *arg1) + const struct brw_reg *arg0) { struct brw_reg r1 = brw_vec1_grf(1, 0); @@ -545,16 +543,18 @@ static void emit_dp3( struct brw_compile *p, const struct brw_reg *arg0, const struct brw_reg *arg1 ) { + int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1; + if (!(mask & WRITEMASK_XYZW)) return; /* Do not emit dead code */ - assert((mask & WRITEMASK_XYZW) == WRITEMASK_X); + assert(is_power_of_two(mask & WRITEMASK_XYZW)); brw_MUL(p, brw_null_reg(), arg0[0], arg1[0]); brw_MAC(p, brw_null_reg(), arg0[1], arg1[1]); brw_set_saturate(p, (mask & SATURATE) ? 1 : 0); - brw_MAC(p, dst[0], arg0[2], arg1[2]); + brw_MAC(p, dst[dst_chan], arg0[2], arg1[2]); brw_set_saturate(p, 0); } @@ -565,17 +565,19 @@ static void emit_dp4( struct brw_compile *p, const struct brw_reg *arg0, const struct brw_reg *arg1 ) { + int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1; + if (!(mask & WRITEMASK_XYZW)) return; /* Do not emit dead code */ - assert((mask & WRITEMASK_XYZW) == WRITEMASK_X); + assert(is_power_of_two(mask & WRITEMASK_XYZW)); brw_MUL(p, brw_null_reg(), arg0[0], arg1[0]); brw_MAC(p, brw_null_reg(), arg0[1], arg1[1]); brw_MAC(p, brw_null_reg(), arg0[2], arg1[2]); brw_set_saturate(p, (mask & SATURATE) ? 1 : 0); - brw_MAC(p, dst[0], arg0[3], arg1[3]); + brw_MAC(p, dst[dst_chan], arg0[3], arg1[3]); brw_set_saturate(p, 0); } @@ -632,18 +634,19 @@ static void emit_math1( struct brw_compile *p, GLuint mask, const struct brw_reg *arg0 ) { + int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1; + if (!(mask & WRITEMASK_XYZW)) return; /* Do not emit dead code */ - //assert((mask & WRITEMASK_XYZW) == WRITEMASK_X || - // function == BRW_MATH_FUNCTION_SINCOS); - + assert(is_power_of_two(mask & WRITEMASK_XYZW)); + brw_MOV(p, brw_message_reg(2), arg0[0]); /* Send two messages to perform all 16 operations: */ brw_math_16(p, - dst[0], + dst[dst_chan], function, (mask & SATURATE) ? BRW_MATH_SATURATE_SATURATE : BRW_MATH_SATURATE_NONE, 2, @@ -659,10 +662,12 @@ static void emit_math2( struct brw_compile *p, const struct brw_reg *arg0, const struct brw_reg *arg1) { + int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1; + if (!(mask & WRITEMASK_XYZW)) return; /* Do not emit dead code */ - assert((mask & WRITEMASK_XYZW) == WRITEMASK_X); + assert(is_power_of_two(mask & WRITEMASK_XYZW)); brw_push_insn_state(p); @@ -681,7 +686,7 @@ static void emit_math2( struct brw_compile *p, */ brw_set_compression_control(p, BRW_COMPRESSION_NONE); brw_math(p, - dst[0], + dst[dst_chan], function, (mask & SATURATE) ? BRW_MATH_SATURATE_SATURATE : BRW_MATH_SATURATE_NONE, 2, @@ -691,7 +696,7 @@ static void emit_math2( struct brw_compile *p, brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF); brw_math(p, - offset(dst[0],1), + offset(dst[dst_chan],1), function, (mask & SATURATE) ? BRW_MATH_SATURATE_SATURATE : BRW_MATH_SATURATE_NONE, 4, @@ -1194,11 +1199,11 @@ void brw_wm_emit( struct brw_wm_compile *c ) /* Generated instructions for calculating triangle interpolants: */ case WM_PIXELXY: - emit_pixel_xy(p, dst, dst_flags, args[0]); + emit_pixel_xy(p, dst, dst_flags); break; case WM_DELTAXY: - emit_delta_xy(p, dst, dst_flags, args[0], args[1]); + emit_delta_xy(p, dst, dst_flags, args[0]); break; case WM_WPOSXY: @@ -1385,4 +1390,13 @@ void brw_wm_emit( struct brw_wm_compile *c ) inst->dst[i]->hw_reg, inst->dst[i]->spill_slot); } + + if (INTEL_DEBUG & DEBUG_WM) { + int i; + + _mesa_printf("wm-native:\n"); + for (i = 0; i < p->nr_insn; i++) + brw_disasm(stderr, &p->store[i]); + _mesa_printf("\n"); + } } diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c index 606baab9675..5dc076a8257 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_fp.c +++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c @@ -226,9 +226,42 @@ static struct prog_instruction * emit_op(struct brw_wm_compile *c, 0, 0, 0, /* tex unit, target, shadow */ src0, src1, src2); } - +/* Many Mesa opcodes produce the same value across all the result channels. + * We'd rather not have to support that splatting in the opcode implementations, + * and brw_wm_pass*.c wants to optimize them out by shuffling references around + * anyway. We can easily get both by emitting the opcode to one channel, and + * then MOVing it to the others, which brw_wm_pass*.c already understands. + */ +static struct prog_instruction *emit_scalar_insn(struct brw_wm_compile *c, + const struct prog_instruction *inst0) +{ + struct prog_instruction *inst; + unsigned int dst_chan; + unsigned int other_channel_mask; + + if (inst0->DstReg.WriteMask == 0) + return NULL; + + dst_chan = _mesa_ffs(inst0->DstReg.WriteMask) - 1; + inst = get_fp_inst(c); + *inst = *inst0; + inst->DstReg.WriteMask = 1 << dst_chan; + + other_channel_mask = inst0->DstReg.WriteMask & ~(1 << dst_chan); + if (other_channel_mask != 0) { + inst = emit_op(c, + OPCODE_MOV, + dst_mask(inst0->DstReg, other_channel_mask), + 0, + src_swizzle1(src_reg_from_dst(inst0->DstReg), dst_chan), + src_undef(), + src_undef()); + } + return inst; +} + /*********************************************************************** * Special instructions for interpolation and other tasks @@ -690,7 +723,7 @@ static void precalc_tex( struct brw_wm_compile *c, /* tmp0 = 1 / tmp1 */ emit_op(c, OPCODE_RCP, - tmp0, + dst_mask(tmp0, WRITEMASK_X), 0, tmp1src, src_undef(), @@ -701,7 +734,7 @@ static void precalc_tex( struct brw_wm_compile *c, tmpcoord, 0, src0, - tmp0src, + src_swizzle1(tmp0src, SWIZZLE_X), src_undef()); release_temp(c, tmp0); @@ -724,7 +757,11 @@ static void precalc_tex( struct brw_wm_compile *c, tmpcoord, 0, inst->SrcReg[0], - scale, + src_swizzle(scale, + SWIZZLE_X, + SWIZZLE_Y, + SWIZZLE_ONE, + SWIZZLE_ONE), src_undef()); coord = src_reg_from_dst(tmpcoord); @@ -1141,9 +1178,11 @@ void brw_wm_pass_fp( struct brw_wm_compile *c ) break; case OPCODE_PRINT: break; - default: - emit_insn(c, inst); + if (brw_wm_is_scalar_result(inst->Opcode)) + emit_scalar_insn(c, inst); + else + emit_insn(c, inst); break; } } diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c index 19f777fe32e..a5b18ec7d76 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c +++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c @@ -10,6 +10,9 @@ enum _subroutine { SUB_NOISE1, SUB_NOISE2, SUB_NOISE3, SUB_NOISE4 }; +static struct brw_reg get_dst_reg(struct brw_wm_compile *c, + const struct prog_instruction *inst, + GLuint component); /** * Determine if the given fragment program uses GLSL features such @@ -22,6 +25,7 @@ GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp) for (i = 0; i < fp->Base.NumInstructions; i++) { const struct prog_instruction *inst = &fp->Base.Instructions[i]; switch (inst->Opcode) { + case OPCODE_ARL: case OPCODE_IF: case OPCODE_ENDIF: case OPCODE_CAL: @@ -130,19 +134,6 @@ static void set_reg(struct brw_wm_compile *c, int file, int index, c->wm_regs[file][index][component].inited = GL_TRUE; } -/** - * Examine instruction's write mask to find index of first component - * enabled for writing. - */ -static int get_scalar_dst_index(const struct prog_instruction *inst) -{ - int i; - for (i = 0; i < 4; i++) - if (inst->DstReg.WriteMask & (1<<i)) - break; - return i; -} - static struct brw_reg alloc_tmp(struct brw_wm_compile *c) { struct brw_reg reg; @@ -402,6 +393,27 @@ static void prealloc_reg(struct brw_wm_compile *c) prealloc_grf(c, 126); prealloc_grf(c, 127); + for (i = 0; i < c->nr_fp_insns; i++) { + const struct prog_instruction *inst = &c->prog_instructions[i]; + struct brw_reg dst[4]; + + switch (inst->Opcode) { + case OPCODE_TEX: + case OPCODE_TXB: + /* Allocate the channels of texture results contiguously, + * since they are written out that way by the sampler unit. + */ + for (j = 0; j < 4; j++) { + dst[j] = get_dst_reg(c, inst, j); + if (j != 0) + assert(dst[j].nr == dst[j - 1].nr + 1); + } + break; + default: + break; + } + } + /* An instruction may reference up to three constants. * They'll be found in these registers. * XXX alloc these on demand! @@ -639,23 +651,6 @@ static void invoke_subroutine( struct brw_wm_compile *c, } } -static void emit_abs( struct brw_wm_compile *c, - const struct prog_instruction *inst) -{ - int i; - struct brw_compile *p = &c->func; - brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF); - for (i = 0; i < 4; i++) { - if (inst->DstReg.WriteMask & (1<<i)) { - struct brw_reg src, dst; - dst = get_dst_reg(c, inst, i); - src = get_src_reg(c, inst, 0, i); - brw_MOV(p, dst, brw_abs(src)); - } - } - brw_set_saturate(p, 0); -} - static void emit_trunc( struct brw_wm_compile *c, const struct prog_instruction *inst) { @@ -1031,12 +1026,20 @@ static void emit_dp3(struct brw_wm_compile *c, struct brw_reg src0[3], src1[3], dst; int i; struct brw_compile *p = &c->func; + GLuint mask = inst->DstReg.WriteMask; + int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1; + + if (!(mask & WRITEMASK_XYZW)) + return; + + assert(is_power_of_two(mask & WRITEMASK_XYZW)); + for (i = 0; i < 3; i++) { src0[i] = get_src_reg(c, inst, 0, i); src1[i] = get_src_reg_imm(c, inst, 1, i); } - dst = get_dst_reg(c, inst, get_scalar_dst_index(inst)); + dst = get_dst_reg(c, inst, dst_chan); brw_MUL(p, brw_null_reg(), src0[0], src1[0]); brw_MAC(p, brw_null_reg(), src0[1], src1[1]); brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0); @@ -1050,11 +1053,19 @@ static void emit_dp4(struct brw_wm_compile *c, struct brw_reg src0[4], src1[4], dst; int i; struct brw_compile *p = &c->func; + GLuint mask = inst->DstReg.WriteMask; + int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1; + + if (!(mask & WRITEMASK_XYZW)) + return; + + assert(is_power_of_two(mask & WRITEMASK_XYZW)); + for (i = 0; i < 4; i++) { src0[i] = get_src_reg(c, inst, 0, i); src1[i] = get_src_reg_imm(c, inst, 1, i); } - dst = get_dst_reg(c, inst, get_scalar_dst_index(inst)); + dst = get_dst_reg(c, inst, dst_chan); brw_MUL(p, brw_null_reg(), src0[0], src1[0]); brw_MAC(p, brw_null_reg(), src0[1], src1[1]); brw_MAC(p, brw_null_reg(), src0[2], src1[2]); @@ -1069,11 +1080,19 @@ static void emit_dph(struct brw_wm_compile *c, struct brw_reg src0[4], src1[4], dst; int i; struct brw_compile *p = &c->func; + GLuint mask = inst->DstReg.WriteMask; + int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1; + + if (!(mask & WRITEMASK_XYZW)) + return; + + assert(is_power_of_two(mask & WRITEMASK_XYZW)); + for (i = 0; i < 4; i++) { src0[i] = get_src_reg(c, inst, 0, i); src1[i] = get_src_reg_imm(c, inst, 1, i); } - dst = get_dst_reg(c, inst, get_scalar_dst_index(inst)); + dst = get_dst_reg(c, inst, dst_chan); brw_MUL(p, brw_null_reg(), src0[0], src1[0]); brw_MAC(p, brw_null_reg(), src0[1], src1[1]); brw_MAC(p, dst, src0[2], src1[2]); @@ -1091,37 +1110,28 @@ static void emit_math1(struct brw_wm_compile *c, const struct prog_instruction *inst, GLuint func) { struct brw_compile *p = &c->func; - struct brw_reg src0, dst, tmp; - const int mark = mark_tmps( c ); - int i; + struct brw_reg src0, dst; + GLuint mask = inst->DstReg.WriteMask; + int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1; + + if (!(mask & WRITEMASK_XYZW)) + return; - tmp = alloc_tmp(c); + assert(is_power_of_two(mask & WRITEMASK_XYZW)); /* Get first component of source register */ + dst = get_dst_reg(c, inst, dst_chan); src0 = get_src_reg(c, inst, 0, 0); - /* tmp = func(src0) */ brw_MOV(p, brw_message_reg(2), src0); brw_math(p, - tmp, + dst, func, (inst->SaturateMode != SATURATE_OFF) ? BRW_MATH_SATURATE_SATURATE : BRW_MATH_SATURATE_NONE, 2, brw_null_reg(), BRW_MATH_DATA_VECTOR, BRW_MATH_PRECISION_FULL); - - /*tmp.dw1.bits.swizzle = SWIZZLE_XXXX;*/ - - /* replicate tmp value across enabled dest channels */ - for (i = 0; i < 4; i++) { - if (inst->DstReg.WriteMask & (1 << i)) { - dst = get_dst_reg(c, inst, i); - brw_MOV(p, dst, tmp); - } - } - - release_tmps(c, mark); } static void emit_rcp(struct brw_wm_compile *c, @@ -1192,24 +1202,6 @@ static void emit_arl(struct brw_wm_compile *c, brw_set_saturate(p, 0); } -static void emit_sub(struct brw_wm_compile *c, - const struct prog_instruction *inst) -{ - struct brw_compile *p = &c->func; - struct brw_reg src0, src1, dst; - GLuint mask = inst->DstReg.WriteMask; - int i; - brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0); - for (i = 0 ; i < 4; i++) { - if (mask & (1<<i)) { - dst = get_dst_reg(c, inst, i); - src0 = get_src_reg(c, inst, 0, i); - src1 = get_src_reg_imm(c, inst, 1, i); - brw_ADD(p, dst, src0, negate(src1)); - } - } - brw_set_saturate(p, 0); -} static void emit_mul(struct brw_wm_compile *c, const struct prog_instruction *inst) @@ -1321,7 +1313,15 @@ static void emit_pow(struct brw_wm_compile *c, { struct brw_compile *p = &c->func; struct brw_reg dst, src0, src1; - dst = get_dst_reg(c, inst, get_scalar_dst_index(inst)); + GLuint mask = inst->DstReg.WriteMask; + int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1; + + if (!(mask & WRITEMASK_XYZW)) + return; + + assert(is_power_of_two(mask & WRITEMASK_XYZW)); + + dst = get_dst_reg(c, inst, dst_chan); src0 = get_src_reg_imm(c, inst, 0, 0); src1 = get_src_reg_imm(c, inst, 1, 0); @@ -2828,18 +2828,12 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c) case WM_FRONTFACING: emit_frontfacing(c, inst); break; - case OPCODE_ABS: - emit_abs(c, inst); - break; case OPCODE_ADD: emit_add(c, inst); break; case OPCODE_ARL: emit_arl(c, inst); break; - case OPCODE_SUB: - emit_sub(c, inst); - break; case OPCODE_FRC: emit_frc(c, inst); break; @@ -3007,7 +3001,7 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c) loop_depth--; inst0 = inst1 = brw_WHILE(p, loop_inst[loop_depth]); - /* patch all the BREAK/CONT instructions from last BEGINLOOP */ + /* patch all the BREAK/CONT instructions from last BGNLOOP */ while (inst0 > loop_inst[loop_depth]) { inst0--; if (inst0->header.opcode == BRW_OPCODE_BREAK) { @@ -3032,8 +3026,14 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c) brw_set_predicate_control(p, BRW_PREDICATE_NONE); } post_wm_emit(c); -} + if (INTEL_DEBUG & DEBUG_WM) { + _mesa_printf("wm-native:\n"); + for (i = 0; i < p->nr_insn; i++) + brw_disasm(stderr, &p->store[i]); + _mesa_printf("\n"); + } +} /** * Do GPU code generation for shaders that use GLSL features such as diff --git a/src/mesa/drivers/dri/i965/brw_wm_iz.c b/src/mesa/drivers/dri/i965/brw_wm_iz.c index 8fd067abe7d..5e399ac62a8 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_iz.c +++ b/src/mesa/drivers/dri/i965/brw_wm_iz.c @@ -122,6 +122,7 @@ const struct { */ void brw_wm_lookup_iz( GLuint line_aa, GLuint lookup, + GLboolean ps_uses_depth, struct brw_wm_prog_key *key ) { GLuint reg = 2; @@ -131,7 +132,7 @@ void brw_wm_lookup_iz( GLuint line_aa, if (lookup & IZ_PS_COMPUTES_DEPTH_BIT) key->computes_depth = 1; - if (wm_iz_table[lookup].sd_present) { + if (wm_iz_table[lookup].sd_present || ps_uses_depth) { key->source_depth_reg = reg; reg += 2; } diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass0.c b/src/mesa/drivers/dri/i965/brw_wm_pass0.c index 92142764f5d..62792583396 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_pass0.c +++ b/src/mesa/drivers/dri/i965/brw_wm_pass0.c @@ -257,34 +257,6 @@ static void pass0_set_dst( struct brw_wm_compile *c, } -static void pass0_set_dst_scalar( struct brw_wm_compile *c, - struct brw_wm_instruction *out, - const struct prog_instruction *inst, - GLuint writemask ) -{ - if (writemask) { - const struct prog_dst_register *dst = &inst->DstReg; - GLuint i; - - /* Compute only the first (X) value: - */ - out->writemask = WRITEMASK_X; - out->dst[0] = get_value(c); - - /* Update our tracking register file for all the components in - * writemask: - */ - for (i = 0; i < 4; i++) { - if (writemask & (1<<i)) { - pass0_set_fpreg_value(c, dst->File, dst->Index, i, out->dst[0]); - } - } - } - else - out->writemask = 0; -} - - static const struct brw_wm_ref *get_fp_src_reg_ref( struct brw_wm_compile *c, struct prog_src_register src, GLuint i ) @@ -363,10 +335,7 @@ translate_insn(struct brw_wm_compile *c, /* Dst: */ - if (brw_wm_is_scalar_result(out->opcode)) - pass0_set_dst_scalar(c, out, inst, writemask); - else - pass0_set_dst(c, out, inst, writemask); + pass0_set_dst(c, out, inst, writemask); } diff --git a/src/mesa/drivers/dri/intel/intel_batchbuffer.c b/src/mesa/drivers/dri/intel/intel_batchbuffer.c index 0f87fc46a44..6aa36d10b12 100644 --- a/src/mesa/drivers/dri/intel/intel_batchbuffer.c +++ b/src/mesa/drivers/dri/intel/intel_batchbuffer.c @@ -196,6 +196,16 @@ _intel_batchbuffer_flush(struct intel_batchbuffer *batch, const char *file, struct intel_context *intel = batch->intel; GLuint used = batch->ptr - batch->map; + if (intel->first_post_swapbuffers_batch == NULL) { + intel->first_post_swapbuffers_batch = intel->batch->buf; + drm_intel_bo_reference(intel->first_post_swapbuffers_batch); + } + + if (intel->first_post_swapbuffers_batch == NULL) { + intel->first_post_swapbuffers_batch = intel->batch->buf; + drm_intel_bo_reference(intel->first_post_swapbuffers_batch); + } + if (used == 0) { batch->cliprect_mode = IGNORE_CLIPRECTS; return; diff --git a/src/mesa/drivers/dri/intel/intel_blit.c b/src/mesa/drivers/dri/intel/intel_blit.c index 2e95bd1013f..979f2025842 100644 --- a/src/mesa/drivers/dri/intel/intel_blit.c +++ b/src/mesa/drivers/dri/intel/intel_blit.c @@ -477,6 +477,8 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask) BR13 |= BR13_565; } + assert(irb->region->tiling != I915_TILING_Y); + #ifndef I915 if (irb->region->tiling != I915_TILING_NONE) { CMD |= XY_DST_TILED; @@ -571,6 +573,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel, assert( logic_op - GL_CLEAR >= 0 ); assert( logic_op - GL_CLEAR < 0x10 ); + assert(dst_pitch > 0); if (w < 0 || h < 0) return GL_TRUE; diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c index 4abb525f78e..46f1a7f7205 100644 --- a/src/mesa/drivers/dri/intel/intel_context.c +++ b/src/mesa/drivers/dri/intel/intel_context.c @@ -38,6 +38,7 @@ #include "swrast_setup/swrast_setup.h" #include "tnl/tnl.h" #include "drivers/common/driverfuncs.h" +#include "drivers/common/meta.h" #include "i830_dri.h" @@ -513,7 +514,7 @@ intel_flush(GLcontext *ctx, GLboolean needs_mi_flush) * each of N places that do rendering. This has worse performances, * but it is much easier to get correct. */ - if (intel->is_front_buffer_rendering) { + if (!intel->is_front_buffer_rendering) { intel->front_buffer_dirty = GL_FALSE; } } @@ -529,7 +530,27 @@ intelFlush(GLcontext * ctx) static void intel_glFlush(GLcontext *ctx) { + struct intel_context *intel = intel_context(ctx); + intel_flush(ctx, GL_TRUE); + + /* We're using glFlush as an indicator that a frame is done, which is + * what DRI2 does before calling SwapBuffers (and means we should catch + * people doing front-buffer rendering, as well).. + * + * Wait for the swapbuffers before the one we just emitted, so we don't + * get too many swaps outstanding for apps that are GPU-heavy but not + * CPU-heavy. + * + * Unfortunately, we don't have a handle to the batch containing the swap, + * and getting our hands on that doesn't seem worth it, so we just us the + * first batch we emitted after the last swap. + */ + if (intel->first_post_swapbuffers_batch != NULL) { + drm_intel_bo_wait_rendering(intel->first_post_swapbuffers_batch); + drm_intel_bo_unreference(intel->first_post_swapbuffers_batch); + intel->first_post_swapbuffers_batch = NULL; + } } void @@ -692,6 +713,8 @@ intelInitContext(struct intel_context *intel, _swrast_allow_pixel_fog(ctx, GL_FALSE); _swrast_allow_vertex_fog(ctx, GL_TRUE); + _mesa_meta_init(ctx); + intel->hw_stencil = mesaVis->stencilBits && mesaVis->depthBits == 24; intel->hw_stipple = 1; @@ -752,7 +775,7 @@ intelInitContext(struct intel_context *intel, if (intel->use_texture_tiling && !intel->intelScreen->kernel_exec_fencing) { fprintf(stderr, "No kernel support for execution fencing, " - "disabling texture tiling"); + "disabling texture tiling\n"); intel->use_texture_tiling = GL_FALSE; } intel->use_early_z = driQueryOptionb(&intel->optionCache, "early_z"); @@ -795,6 +818,8 @@ intelDestroyContext(__DRIcontextPrivate * driContextPriv) INTEL_FIREVERTICES(intel); + _mesa_meta_free(&intel->ctx); + meta_destroy_metaops(&intel->meta); intel->vtbl.destroy(intel); @@ -814,6 +839,8 @@ intelDestroyContext(__DRIcontextPrivate * driContextPriv) intel->prim.vb = NULL; dri_bo_unreference(intel->prim.vb_bo); intel->prim.vb_bo = NULL; + dri_bo_unreference(intel->first_post_swapbuffers_batch); + intel->first_post_swapbuffers_batch = NULL; if (release_texture_heaps) { /* Nothing is currently done here to free texture heaps; diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h index 08bea88c958..2b138290821 100644 --- a/src/mesa/drivers/dri/intel/intel_context.h +++ b/src/mesa/drivers/dri/intel/intel_context.h @@ -80,9 +80,13 @@ extern void intelFallback(struct intel_context *intel, GLuint bit, #define INTEL_MAX_FIXUP 64 + +/** + * intel_context is derived from Mesa's context class: GLcontext. + */ struct intel_context { - GLcontext ctx; /* the parent class */ + GLcontext ctx; /**< base class, must be first field */ struct { @@ -178,6 +182,7 @@ struct intel_context GLboolean ttm; struct intel_batchbuffer *batch; + drm_intel_bo *first_post_swapbuffers_batch; GLboolean no_batch_wrap; unsigned batch_id; diff --git a/src/mesa/drivers/dri/intel/intel_extensions.c b/src/mesa/drivers/dri/intel/intel_extensions.c index 7742609d242..aa3d704f299 100644 --- a/src/mesa/drivers/dri/intel/intel_extensions.c +++ b/src/mesa/drivers/dri/intel/intel_extensions.c @@ -48,6 +48,7 @@ #define need_GL_EXT_framebuffer_blit #define need_GL_EXT_gpu_program_parameters #define need_GL_EXT_point_parameters +#define need_GL_EXT_provoking_vertex #define need_GL_EXT_secondary_color #define need_GL_EXT_stencil_two_side #define need_GL_APPLE_vertex_array_object @@ -71,6 +72,7 @@ static const struct dri_extension card_extensions[] = { { "GL_ARB_half_float_pixel", NULL }, { "GL_ARB_multitexture", NULL }, { "GL_ARB_point_parameters", GL_ARB_point_parameters_functions }, + { "GL_ARB_point_sprite", NULL }, { "GL_ARB_texture_border_clamp", NULL }, { "GL_ARB_texture_cube_map", NULL }, { "GL_ARB_texture_env_add", NULL }, @@ -92,6 +94,7 @@ static const struct dri_extension card_extensions[] = { { "GL_EXT_fog_coord", GL_EXT_fog_coord_functions }, { "GL_EXT_gpu_program_parameters", GL_EXT_gpu_program_parameters_functions }, { "GL_EXT_packed_depth_stencil", NULL }, + { "GL_EXT_provoking_vertex", GL_EXT_provoking_vertex_functions }, { "GL_EXT_secondary_color", GL_EXT_secondary_color_functions }, { "GL_EXT_stencil_wrap", NULL }, { "GL_EXT_texture_edge_clamp", NULL }, @@ -119,8 +122,10 @@ static const struct dri_extension i915_extensions[] = { { "GL_ARB_fragment_program", NULL }, { "GL_ARB_shadow", NULL }, { "GL_ARB_texture_non_power_of_two", NULL }, + { "GL_ATI_separate_stencil", GL_ATI_separate_stencil_functions }, { "GL_ATI_texture_env_combine3", NULL }, { "GL_EXT_shadow_funcs", NULL }, + { "GL_EXT_stencil_two_side", GL_EXT_stencil_two_side_functions }, { "GL_NV_texture_env_combine4", NULL }, { NULL, NULL } }; diff --git a/src/mesa/drivers/dri/intel/intel_fbo.c b/src/mesa/drivers/dri/intel/intel_fbo.c index 666893596e1..804c0348401 100644 --- a/src/mesa/drivers/dri/intel/intel_fbo.c +++ b/src/mesa/drivers/dri/intel/intel_fbo.c @@ -35,6 +35,7 @@ #include "main/context.h" #include "main/texformat.h" #include "main/texrender.h" +#include "drivers/common/meta.h" #include "intel_context.h" #include "intel_buffers.h" @@ -700,74 +701,6 @@ intel_validate_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb) /** - * Called from glBlitFramebuffer(). - * For now, we're doing an approximation with glCopyPixels(). - * XXX we need to bypass all the per-fragment operations, except scissor. - */ -static void -intel_blit_framebuffer(GLcontext *ctx, - GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, - GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, - GLbitfield mask, GLenum filter) -{ - const GLfloat xZoomSave = ctx->Pixel.ZoomX; - const GLfloat yZoomSave = ctx->Pixel.ZoomY; - GLsizei width, height; - GLfloat xFlip = 1.0F, yFlip = 1.0F; - - if (srcX1 < srcX0) { - GLint tmp = srcX1; - srcX1 = srcX0; - srcX0 = tmp; - xFlip = -1.0F; - } - - if (srcY1 < srcY0) { - GLint tmp = srcY1; - srcY1 = srcY0; - srcY0 = tmp; - yFlip = -1.0F; - } - - width = srcX1 - srcX0; - height = srcY1 - srcY0; - - ctx->Pixel.ZoomX = xFlip * (dstX1 - dstX0) / (srcX1 - srcY0); - ctx->Pixel.ZoomY = yFlip * (dstY1 - dstY0) / (srcY1 - srcY0); - - if (ctx->Pixel.ZoomX < 0.0F) { - dstX0 = MAX2(dstX0, dstX1); - } - else { - dstX0 = MIN2(dstX0, dstX1); - } - - if (ctx->Pixel.ZoomY < 0.0F) { - dstY0 = MAX2(dstY0, dstY1); - } - else { - dstY0 = MIN2(dstY0, dstY1); - } - - if (mask & GL_COLOR_BUFFER_BIT) { - ctx->Driver.CopyPixels(ctx, srcX0, srcY0, width, height, - dstX0, dstY0, GL_COLOR); - } - if (mask & GL_DEPTH_BUFFER_BIT) { - ctx->Driver.CopyPixels(ctx, srcX0, srcY0, width, height, - dstX0, dstY0, GL_DEPTH); - } - if (mask & GL_STENCIL_BUFFER_BIT) { - ctx->Driver.CopyPixels(ctx, srcX0, srcY0, width, height, - dstX0, dstY0, GL_STENCIL); - } - - ctx->Pixel.ZoomX = xZoomSave; - ctx->Pixel.ZoomY = yZoomSave; -} - - -/** * Do one-time context initializations related to GL_EXT_framebuffer_object. * Hook in device driver functions. */ @@ -782,5 +715,5 @@ intel_fbo_init(struct intel_context *intel) intel->ctx.Driver.FinishRenderTexture = intel_finish_render_texture; intel->ctx.Driver.ResizeBuffers = intel_resize_buffers; intel->ctx.Driver.ValidateFramebuffer = intel_validate_framebuffer; - intel->ctx.Driver.BlitFramebuffer = intel_blit_framebuffer; + intel->ctx.Driver.BlitFramebuffer = _mesa_meta_blit_framebuffer; } diff --git a/src/mesa/drivers/dri/intel/intel_pixel_copy.c b/src/mesa/drivers/dri/intel/intel_pixel_copy.c index 5d52335dee3..ca796b36559 100644 --- a/src/mesa/drivers/dri/intel/intel_pixel_copy.c +++ b/src/mesa/drivers/dri/intel/intel_pixel_copy.c @@ -31,7 +31,7 @@ #include "main/state.h" #include "main/mtypes.h" #include "main/macros.h" -#include "swrast/swrast.h" +#include "drivers/common/meta.h" #include "intel_screen.h" #include "intel_context.h" @@ -97,162 +97,6 @@ intel_check_copypixel_blit_fragment_ops(GLcontext * ctx) ctx->Color.BlendEnabled); } -#ifdef I915 -/* Doesn't work for overlapping regions. Could do a double copy or - * just fallback. - */ -static GLboolean -do_texture_copypixels(GLcontext * ctx, - GLint srcx, GLint srcy, - GLsizei width, GLsizei height, - GLint dstx, GLint dsty, GLenum type) -{ - struct intel_context *intel = intel_context(ctx); - struct intel_region *dst = intel_drawbuf_region(intel); - struct intel_region *src = copypix_src_region(intel, type); - GLenum src_format; - GLenum src_type; - - DBG("%s %d,%d %dx%d --> %d,%d\n", __FUNCTION__, - srcx, srcy, width, height, dstx, dsty); - - if (!src || !dst || type != GL_COLOR) - return GL_FALSE; - - if (ctx->_ImageTransferState) { - if (INTEL_DEBUG & DEBUG_PIXEL) - fprintf(stderr, "%s: check_color failed\n", __FUNCTION__); - return GL_FALSE; - } - - /* Can't handle overlapping regions. Don't have sufficient control - * over rasterization to pull it off in-place. Punt on these for - * now. - * - * XXX: do a copy to a temporary. - */ - if (src->buffer == dst->buffer) { - drm_clip_rect_t srcbox; - drm_clip_rect_t dstbox; - drm_clip_rect_t tmp; - - srcbox.x1 = srcx; - srcbox.y1 = srcy; - srcbox.x2 = srcx + width; - srcbox.y2 = srcy + height; - - if (ctx->Pixel.ZoomX > 0) { - dstbox.x1 = dstx; - dstbox.x2 = dstx + width * ctx->Pixel.ZoomX; - } else { - dstbox.x1 = dstx + width * ctx->Pixel.ZoomX; - dstbox.x2 = dstx; - } - if (ctx->Pixel.ZoomY > 0) { - dstbox.y1 = dsty; - dstbox.y2 = dsty + height * ctx->Pixel.ZoomY; - } else { - dstbox.y1 = dsty + height * ctx->Pixel.ZoomY; - dstbox.y2 = dsty; - } - - DBG("src %d,%d %d,%d\n", srcbox.x1, srcbox.y1, srcbox.x2, srcbox.y2); - DBG("dst %d,%d %d,%d (%dx%d) (%f,%f)\n", dstbox.x1, dstbox.y1, dstbox.x2, dstbox.y2, - width, height, ctx->Pixel.ZoomX, ctx->Pixel.ZoomY); - - if (intel_intersect_cliprects(&tmp, &srcbox, &dstbox)) { - DBG("%s: regions overlap\n", __FUNCTION__); - return GL_FALSE; - } - } - - intelFlush(&intel->ctx); - - intel->vtbl.install_meta_state(intel); - - /* Is this true? Also will need to turn depth testing on according - * to state: - */ - intel->vtbl.meta_no_stencil_write(intel); - intel->vtbl.meta_no_depth_write(intel); - - /* Set the 3d engine to draw into the destination region: - */ - intel->vtbl.meta_draw_region(intel, dst, intel->depth_region); - - intel->vtbl.meta_import_pixel_state(intel); - - if (src->cpp == 2) { - src_format = GL_RGB; - src_type = GL_UNSIGNED_SHORT_5_6_5; - } - else { - src_format = GL_BGRA; - src_type = GL_UNSIGNED_BYTE; - } - - /* Set the frontbuffer up as a large rectangular texture. - */ - if (!intel->vtbl.meta_tex_rect_source(intel, src->buffer, 0, - src->pitch, - src->height, src_format, src_type)) { - intel->vtbl.leave_meta_state(intel); - return GL_FALSE; - } - - - intel->vtbl.meta_texture_blend_replace(intel); - - LOCK_HARDWARE(intel); - - if (intel->driDrawable->numClipRects) { - __DRIdrawablePrivate *dPriv = intel->driDrawable; - - - srcy = dPriv->h - srcy - height; /* convert from gl to hardware coords */ - - srcx += dPriv->x; - srcy += dPriv->y; - - /* Clip against the source region. This is the only source - * clipping we do. XXX: Just set the texcord wrap mode to clamp - * or similar. - * - */ - if (0) { - GLint orig_x = srcx; - GLint orig_y = srcy; - - if (!_mesa_clip_to_region(0, 0, src->pitch, src->height, - &srcx, &srcy, &width, &height)) - goto out; - - dstx += srcx - orig_x; - dsty += (srcy - orig_y) * ctx->Pixel.ZoomY; - } - - /* Just use the regular cliprect mechanism... Does this need to - * even hold the lock??? - */ - intel->vtbl.meta_draw_quad(intel, - dstx, - dstx + width * ctx->Pixel.ZoomX, - dPriv->h - (dsty + height * ctx->Pixel.ZoomY), - dPriv->h - (dsty), 0, /* XXX: what z value? */ - 0x00ff00ff, - srcx, srcx + width, srcy, srcy + height); - - out: - intel->vtbl.leave_meta_state(intel); - intel_batchbuffer_emit_mi_flush(intel->batch); - } - UNLOCK_HARDWARE(intel); - - DBG("%s: success\n", __FUNCTION__); - return GL_TRUE; -} -#endif /* I915 */ - /** * CopyPixels with the blitter. Don't support zooming, pixel transfer, etc. @@ -400,12 +244,5 @@ intelCopyPixels(GLcontext * ctx, if (do_blit_copypixels(ctx, srcx, srcy, width, height, destx, desty, type)) return; -#ifdef I915 - if (do_texture_copypixels(ctx, srcx, srcy, width, height, destx, desty, type)) - return; -#endif - - DBG("fallback to _swrast_CopyPixels\n"); - - _swrast_CopyPixels(ctx, srcx, srcy, width, height, destx, desty, type); + _mesa_meta_copy_pixels(ctx, srcx, srcy, width, height, destx, desty, type); } diff --git a/src/mesa/drivers/dri/intel/intel_regions.c b/src/mesa/drivers/dri/intel/intel_regions.c index 7525cd9c4d7..497f7967649 100644 --- a/src/mesa/drivers/dri/intel/intel_regions.c +++ b/src/mesa/drivers/dri/intel/intel_regions.c @@ -181,6 +181,11 @@ intel_region_alloc(struct intel_context *intel, dri_bo *buffer; struct intel_region *region; + if (tiling == I915_TILING_X) + height = ALIGN(height, 8); + else if (tiling == I915_TILING_Y) + height = ALIGN(height, 32); + if (expect_accelerated_upload) { buffer = drm_intel_bo_alloc_for_render(intel->bufmgr, "region", pitch * cpp * height, 64); @@ -452,6 +457,7 @@ void intel_region_cow(struct intel_context *intel, struct intel_region *region) { struct intel_buffer_object *pbo = region->pbo; + GLboolean ok; intel_region_release_pbo(intel, region); @@ -463,13 +469,14 @@ intel_region_cow(struct intel_context *intel, struct intel_region *region) */ LOCK_HARDWARE(intel); - assert(intelEmitCopyBlit(intel, - region->cpp, - region->pitch, pbo->buffer, 0, region->tiling, - region->pitch, region->buffer, 0, region->tiling, - 0, 0, 0, 0, - region->pitch, region->height, - GL_COPY)); + ok = intelEmitCopyBlit(intel, + region->cpp, + region->pitch, pbo->buffer, 0, region->tiling, + region->pitch, region->buffer, 0, region->tiling, + 0, 0, 0, 0, + region->pitch, region->height, + GL_COPY); + assert(ok); UNLOCK_HARDWARE(intel); } diff --git a/src/mesa/drivers/dri/intel/intel_screen.c b/src/mesa/drivers/dri/intel/intel_screen.c index 14cc815139c..1b8c56e68d6 100644 --- a/src/mesa/drivers/dri/intel/intel_screen.c +++ b/src/mesa/drivers/dri/intel/intel_screen.c @@ -69,7 +69,11 @@ PUBLIC const char __driConfigOptions[] = DRI_CONF_DESC_END DRI_CONF_OPT_END - DRI_CONF_TEXTURE_TILING(false) +#ifdef I915 + DRI_CONF_TEXTURE_TILING(false) +#else + DRI_CONF_TEXTURE_TILING(true) +#endif DRI_CONF_OPT_BEGIN(early_z, bool, false) DRI_CONF_DESC(en, "Enable early Z in classic mode (unstable, 945-only).") diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c index 34b78ebc1ab..8df49908806 100644 --- a/src/mesa/drivers/dri/intel/intel_span.c +++ b/src/mesa/drivers/dri/intel/intel_span.c @@ -501,7 +501,7 @@ intel_map_unmap_buffers(struct intel_context *intel, GLboolean map) /** - * Prepare for softare rendering. Map current read/draw framebuffers' + * Prepare for software rendering. Map current read/draw framebuffers' * renderbuffes and all currently bound texture objects. * * Old note: Moved locking out to get reasonable span performance. @@ -526,7 +526,7 @@ intelSpanRenderStart(GLcontext * ctx) } /** - * Called when done softare rendering. Unmap the buffers we mapped in + * Called when done software rendering. Unmap the buffers we mapped in * the above function. */ void diff --git a/src/mesa/drivers/dri/intel/intel_tex_subimage.c b/src/mesa/drivers/dri/intel/intel_tex_subimage.c index 1f27131dac0..89037073f84 100644 --- a/src/mesa/drivers/dri/intel/intel_tex_subimage.c +++ b/src/mesa/drivers/dri/intel/intel_tex_subimage.c @@ -44,10 +44,12 @@ intelTexSubimage(GLcontext * ctx, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint width, GLint height, GLint depth, + GLsizei imageSize, GLenum format, GLenum type, const void *pixels, const struct gl_pixelstore_attrib *packing, struct gl_texture_object *texObj, - struct gl_texture_image *texImage) + struct gl_texture_image *texImage, + GLboolean compressed) { struct intel_context *intel = intel_context(ctx); struct intel_texture_image *intelImage = intel_texture_image(texImage); @@ -59,9 +61,14 @@ intelTexSubimage(GLcontext * ctx, intelFlush(ctx); - pixels = - _mesa_validate_pbo_teximage(ctx, dims, width, height, depth, format, - type, pixels, packing, "glTexSubImage2D"); + if (compressed) + pixels = _mesa_validate_pbo_compressed_teximage(ctx, imageSize, + pixels, packing, + "glCompressedTexImage"); + else + pixels = _mesa_validate_pbo_teximage(ctx, dims, width, height, depth, + format, type, pixels, packing, + "glTexSubImage"); if (!pixels) return; @@ -90,15 +97,28 @@ intelTexSubimage(GLcontext * ctx, assert(dstRowStride); - if (!texImage->TexFormat->StoreImage(ctx, dims, texImage->_BaseFormat, - texImage->TexFormat, - texImage->Data, - xoffset, yoffset, zoffset, - dstRowStride, - texImage->ImageOffsets, - width, height, depth, - format, type, pixels, packing)) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "intelTexSubImage"); + if (compressed) { + if (intelImage->mt) { + struct intel_region *dst = intelImage->mt->region; + + _mesa_copy_rect(texImage->Data, dst->cpp, dst->pitch, + xoffset, yoffset / 4, + (width + 3) & ~3, (height + 3) / 4, + pixels, (width + 3) & ~3, 0, 0); + } else + memcpy(texImage->Data, pixels, imageSize); + } + else { + if (!texImage->TexFormat->StoreImage(ctx, dims, texImage->_BaseFormat, + texImage->TexFormat, + texImage->Data, + xoffset, yoffset, zoffset, + dstRowStride, + texImage->ImageOffsets, + width, height, depth, + format, type, pixels, packing)) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "intelTexSubImage"); + } } _mesa_unmap_teximage_pbo(ctx, packing); @@ -132,8 +152,8 @@ intelTexSubImage3D(GLcontext * ctx, intelTexSubimage(ctx, 3, target, level, xoffset, yoffset, zoffset, - width, height, depth, - format, type, pixels, packing, texObj, texImage); + width, height, depth, 0, + format, type, pixels, packing, texObj, texImage, GL_FALSE); } @@ -152,8 +172,8 @@ intelTexSubImage2D(GLcontext * ctx, intelTexSubimage(ctx, 2, target, level, xoffset, yoffset, 0, - width, height, 1, - format, type, pixels, packing, texObj, texImage); + width, height, 1, 0, + format, type, pixels, packing, texObj, texImage, GL_FALSE); } @@ -172,8 +192,8 @@ intelTexSubImage1D(GLcontext * ctx, intelTexSubimage(ctx, 1, target, level, xoffset, 0, 0, - width, 1, 1, - format, type, pixels, packing, texObj, texImage); + width, 1, 1, 0, + format, type, pixels, packing, texObj, texImage, GL_FALSE); } static void @@ -187,8 +207,11 @@ intelCompressedTexSubImage2D(GLcontext * ctx, struct gl_texture_object *texObj, struct gl_texture_image *texImage) { - fprintf(stderr, "stubbed CompressedTexSubImage2D: %dx%d@%dx%d\n", - width, height, xoffset, yoffset); + intelTexSubimage(ctx, 2, + target, level, + xoffset, yoffset, 0, + width, height, 1, imageSize, + format, 0, pixels, &ctx->Unpack, texObj, texImage, GL_TRUE); } diff --git a/src/mesa/drivers/dri/r200/r200_cmdbuf.c b/src/mesa/drivers/dri/r200/r200_cmdbuf.c index d49f4fabe72..5d0f367b387 100644 --- a/src/mesa/drivers/dri/r200/r200_cmdbuf.c +++ b/src/mesa/drivers/dri/r200/r200_cmdbuf.c @@ -107,31 +107,37 @@ void r200SetUpAtomList( r200ContextPtr rmesa ) void r200EmitScissor(r200ContextPtr rmesa) { + unsigned x1, y1, x2, y2; + struct radeon_renderbuffer *rrb; BATCH_LOCALS(&rmesa->radeon); if (!rmesa->radeon.radeonScreen->kernel_mm) { return; } + rrb = radeon_get_colorbuffer(&rmesa->radeon); + if (!rrb || !rrb->bo) + return; + if (rmesa->radeon.state.scissor.enabled) { - BEGIN_BATCH(8); - OUT_BATCH(CP_PACKET0(R200_RE_CNTL, 0)); - OUT_BATCH(R200_SCISSOR_ENABLE | rmesa->hw.set.cmd[SET_RE_CNTL]); - OUT_BATCH(CP_PACKET0(R200_RE_AUX_SCISSOR_CNTL, 0)); - OUT_BATCH(R200_SCISSOR_ENABLE_0); - OUT_BATCH(CP_PACKET0(R200_RE_SCISSOR_TL_0, 0)); - OUT_BATCH((rmesa->radeon.state.scissor.rect.y1 << 16) | - rmesa->radeon.state.scissor.rect.x1); - OUT_BATCH(CP_PACKET0(R200_RE_SCISSOR_BR_0, 0)); - OUT_BATCH(((rmesa->radeon.state.scissor.rect.y2 - 1) << 16) | - (rmesa->radeon.state.scissor.rect.x2 - 1)); - END_BATCH(); + x1 = rmesa->radeon.state.scissor.rect.x1; + y1 = rmesa->radeon.state.scissor.rect.y1; + x2 = rmesa->radeon.state.scissor.rect.x2 - 1; + y2 = rmesa->radeon.state.scissor.rect.y2 - 1; } else { - BEGIN_BATCH(4); - OUT_BATCH(CP_PACKET0(R200_RE_CNTL, 0)); - OUT_BATCH(rmesa->hw.set.cmd[SET_RE_CNTL] & ~R200_SCISSOR_ENABLE); - OUT_BATCH(CP_PACKET0(R200_RE_AUX_SCISSOR_CNTL, 0)); - OUT_BATCH(0); - END_BATCH(); + x1 = 0; + y1 = 0; + x2 = rrb->base.Width - 1; + y2 = rrb->base.Height - 1; } + BEGIN_BATCH(8); + OUT_BATCH(CP_PACKET0(R200_RE_CNTL, 0)); + OUT_BATCH(R200_SCISSOR_ENABLE | rmesa->hw.set.cmd[SET_RE_CNTL]); + OUT_BATCH(CP_PACKET0(R200_RE_AUX_SCISSOR_CNTL, 0)); + OUT_BATCH(0); + OUT_BATCH(CP_PACKET0(R200_RE_TOP_LEFT, 0)); + OUT_BATCH((y1 << 16) | x1); + OUT_BATCH(CP_PACKET0(R200_RE_WIDTH_HEIGHT, 0)); + OUT_BATCH((y2 << 16) | x2); + END_BATCH(); } /* Fire a section of the retained (indexed_verts) buffer as a regular diff --git a/src/mesa/drivers/dri/r200/r200_context.c b/src/mesa/drivers/dri/r200/r200_context.c index 9a92a320797..8cb287de268 100644 --- a/src/mesa/drivers/dri/r200/r200_context.c +++ b/src/mesa/drivers/dri/r200/r200_context.c @@ -500,3 +500,15 @@ GLboolean r200CreateContext( const __GLcontextModes *glVisual, } +void r200DestroyContext( __DRIcontextPrivate *driContextPriv ) +{ + int i; + r200ContextPtr rmesa = (r200ContextPtr)driContextPriv->driverPrivate; + if (rmesa) + { + for ( i = 0 ; i < R200_MAX_TEXTURE_UNITS ; i++ ) { + _math_matrix_dtr( &rmesa->TexGenMatrix[i] ); + } + } + radeonDestroyContext(driContextPriv); +} diff --git a/src/mesa/drivers/dri/r200/r200_state_init.c b/src/mesa/drivers/dri/r200/r200_state_init.c index bc871d99048..78ad5baebb3 100644 --- a/src/mesa/drivers/dri/r200/r200_state_init.c +++ b/src/mesa/drivers/dri/r200/r200_state_init.c @@ -515,7 +515,7 @@ static void ctx_emit_cs(GLcontext *ctx, struct radeon_state_atom *atom) if (drb) dwords += 6; if (rrb) - dwords += 6; + dwords += 8; if (atom->cmd_size == CTX_STATE_SIZE_NEWDRM) dwords += 4; @@ -546,7 +546,7 @@ static void ctx_emit_cs(GLcontext *ctx, struct radeon_state_atom *atom) OUT_BATCH_RELOC(0, rrb->bo, 0, 0, RADEON_GEM_DOMAIN_VRAM, 0); OUT_BATCH(CP_PACKET0(RADEON_RB3D_COLORPITCH, 0)); - OUT_BATCH(cbpitch); + OUT_BATCH_RELOC(cbpitch, rrb->bo, cbpitch, 0, RADEON_GEM_DOMAIN_VRAM, 0); } if (atom->cmd_size == CTX_STATE_SIZE_NEWDRM) { @@ -563,7 +563,6 @@ static void tex_emit(GLcontext *ctx, struct radeon_state_atom *atom) uint32_t dwords = atom->cmd_size; int i = atom->idx; radeonTexObj *t = r200->state.texture.unit[i].texobj; - radeon_mipmap_level *lvl; if (t && t->mt && !t->image_override) dwords += 2; @@ -591,7 +590,6 @@ static void tex_emit_cs(GLcontext *ctx, struct radeon_state_atom *atom) uint32_t dwords = atom->cmd_size; int i = atom->idx; radeonTexObj *t = r200->state.texture.unit[i].texobj; - radeon_mipmap_level *lvl; int hastexture = 1; if (!r200->state.texture.unit[i].unitneeded) @@ -774,7 +772,7 @@ void r200InitState( r200ContextPtr rmesa ) ALLOC_STATE( afs[1], never, AFS_STATE_SIZE, "AFS/afsinst-1", 1 ); } - for (i = 0; i < 5; i++) + for (i = 0; i < 6; i++) if (rmesa->radeon.radeonScreen->kernel_mm) rmesa->hw.tex[i].emit = tex_emit_cs; else @@ -786,7 +784,7 @@ void r200InitState( r200ContextPtr rmesa ) ALLOC_STATE( cube[3], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-3", 3 ); ALLOC_STATE( cube[4], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-4", 4 ); ALLOC_STATE( cube[5], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-5", 5 ); - for (i = 0; i < 5; i++) + for (i = 0; i < 6; i++) if (rmesa->radeon.radeonScreen->kernel_mm) rmesa->hw.cube[i].emit = cube_emit_cs; else diff --git a/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c b/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c index 3a527210c17..d694725c9bb 100644 --- a/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c +++ b/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c @@ -279,7 +279,6 @@ void r500BuildFragmentProgramHwCode(struct r300_fragment_program_compiler *compi _mesa_bzero(code, sizeof(*code)); code->max_temp_idx = 1; - code->inst_offset = 0; code->inst_end = -1; radeonPairProgram(compiler, &pair_handler, compiler); diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_code.h b/src/mesa/drivers/dri/r300/compiler/radeon_code.h index 6f5bc288316..0806fb1b5c6 100644 --- a/src/mesa/drivers/dri/r300/compiler/radeon_code.h +++ b/src/mesa/drivers/dri/r300/compiler/radeon_code.h @@ -160,8 +160,7 @@ struct r500_fragment_program_code { uint32_t inst5; } inst[R500_PFS_MAX_INST]; - int inst_offset; - int inst_end; + int inst_end; /* Number of instructions - 1; also, last instruction to be executed */ int max_temp_idx; }; diff --git a/src/mesa/drivers/dri/r300/r300_cmdbuf.c b/src/mesa/drivers/dri/r300/r300_cmdbuf.c index af535037d06..bd46f9acf2e 100644 --- a/src/mesa/drivers/dri/r300/r300_cmdbuf.c +++ b/src/mesa/drivers/dri/r300/r300_cmdbuf.c @@ -164,47 +164,46 @@ static void emit_tex_offsets(GLcontext *ctx, struct radeon_state_atom * atom) r300ContextPtr r300 = R300_CONTEXT(ctx); BATCH_LOCALS(&r300->radeon); int numtmus = packet0_count(r300, r300->hw.tex.offset.cmd); - int notexture = 0; - - if (numtmus) { - int i; - - for(i = 0; i < numtmus; ++i) { - radeonTexObj *t = r300->hw.textures[i]; - - if (!t) - notexture = 1; - } - - if (r300->radeon.radeonScreen->kernel_mm && notexture) { - return; - } - for(i = 0; i < numtmus; ++i) { - radeonTexObj *t = r300->hw.textures[i]; - if (t && !t->image_override) { - BEGIN_BATCH_NO_AUTOSTATE(4); - OUT_BATCH_REGSEQ(R300_TX_OFFSET_0 + (i * 4), 1); - OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, 0, - RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0); - END_BATCH(); - } else if (!t) { - /* Texture unit hasn't a texture bound nothings to do */ - } else { /* override cases */ - if (t->bo) { - BEGIN_BATCH_NO_AUTOSTATE(4); - OUT_BATCH_REGSEQ(R300_TX_OFFSET_0 + (i * 4), 1); - OUT_BATCH_RELOC(t->tile_bits, t->bo, 0, - RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0); - END_BATCH(); - } else if (!r300->radeon.radeonScreen->kernel_mm) { - BEGIN_BATCH_NO_AUTOSTATE(2); - OUT_BATCH_REGSEQ(R300_TX_OFFSET_0 + (i * 4), 1); - OUT_BATCH(t->override_offset); - END_BATCH(); - } else { - /* Texture unit hasn't a texture bound nothings to do */ - } - } + int i; + + for(i = 0; i < numtmus; ++i) { + radeonTexObj *t = r300->hw.textures[i]; + if (t && !t->image_override) { + BEGIN_BATCH_NO_AUTOSTATE(4); + OUT_BATCH_REGSEQ(R300_TX_OFFSET_0 + (i * 4), 1); + OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, 0, + RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0); + END_BATCH(); + } else if (!t) { + /* Texture unit hasn't a texture bound. + * We assign the current color buffer as a fakery to make + * KIL work on KMS (without it, the CS checker will complain). + */ + if (r300->radeon.radeonScreen->kernel_mm) { + struct radeon_renderbuffer *rrb = radeon_get_colorbuffer(&r300->radeon); + if (rrb && rrb->bo) { + BEGIN_BATCH_NO_AUTOSTATE(4); + OUT_BATCH_REGSEQ(R300_TX_OFFSET_0 + (i * 4), 1); + OUT_BATCH_RELOC(0, rrb->bo, 0, + RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0); + END_BATCH(); + } + } + } else { /* override cases */ + if (t->bo) { + BEGIN_BATCH_NO_AUTOSTATE(4); + OUT_BATCH_REGSEQ(R300_TX_OFFSET_0 + (i * 4), 1); + OUT_BATCH_RELOC(t->tile_bits, t->bo, 0, + RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0); + END_BATCH(); + } else if (!r300->radeon.radeonScreen->kernel_mm) { + BEGIN_BATCH_NO_AUTOSTATE(2); + OUT_BATCH_REGSEQ(R300_TX_OFFSET_0 + (i * 4), 1); + OUT_BATCH(t->override_offset); + END_BATCH(); + } else { + /* Texture unit hasn't a texture bound nothings to do */ + } } } } diff --git a/src/mesa/drivers/dri/r300/r300_context.c b/src/mesa/drivers/dri/r300/r300_context.c index 6f3aab986d2..db404b38479 100644 --- a/src/mesa/drivers/dri/r300/r300_context.c +++ b/src/mesa/drivers/dri/r300/r300_context.c @@ -451,11 +451,6 @@ GLboolean r300CreateContext(const __GLcontextModes * glVisual, r300InitState(r300); r300InitShaderFunctions(r300); - if (screen->chip_family == CHIP_FAMILY_RS600 || screen->chip_family == CHIP_FAMILY_RS690 || - screen->chip_family == CHIP_FAMILY_RS740) { - r300->radeon.texture_row_align = 64; - } - r300InitGLExtensions(ctx); return GL_TRUE; diff --git a/src/mesa/drivers/dri/r300/r300_draw.c b/src/mesa/drivers/dri/r300/r300_draw.c index fcfd3099332..aedc6cfb2a1 100644 --- a/src/mesa/drivers/dri/r300/r300_draw.c +++ b/src/mesa/drivers/dri/r300/r300_draw.c @@ -462,6 +462,7 @@ static void r300DrawPrims(GLcontext *ctx, const struct _mesa_prim *prim, GLuint nr_prims, const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, GLuint min_index, GLuint max_index) { @@ -476,6 +477,12 @@ static void r300DrawPrims(GLcontext *ctx, limits.max_indices = 65535; limits.max_vb_size = 1024*1024; + /* This check should get folded into just the places that + * min/max index are really needed. + */ + if (!index_bounds_valid) + vbo_get_minmax_index(ctx, prim, ib, &min_index, &max_index); + if (min_index) { vbo_rebase_prims( ctx, arrays, prim, nr_prims, ib, min_index, max_index, r300DrawPrims ); return; diff --git a/src/mesa/drivers/dri/r300/r300_ioctl.c b/src/mesa/drivers/dri/r300/r300_ioctl.c index 5bded642ef8..7558f9e225b 100644 --- a/src/mesa/drivers/dri/r300/r300_ioctl.c +++ b/src/mesa/drivers/dri/r300/r300_ioctl.c @@ -168,18 +168,21 @@ static void r300ClearBuffer(r300ContextPtr r300, int flags, } #if 1 if (flags & (CLEARBUFFER_DEPTH | CLEARBUFFER_STENCIL)) { - assert(rrbd != 0); - cbpitch = (rrbd->pitch / rrbd->cpp); + uint32_t zbpitch = (rrbd->pitch / rrbd->cpp); if (rrbd->bo->flags & RADEON_BO_FLAGS_MACRO_TILE){ - cbpitch |= R300_DEPTHMACROTILE_ENABLE; + zbpitch |= R300_DEPTHMACROTILE_ENABLE; } if (rrbd->bo->flags & RADEON_BO_FLAGS_MICRO_TILE){ - cbpitch |= R300_DEPTHMICROTILE_TILED; + zbpitch |= R300_DEPTHMICROTILE_TILED; } BEGIN_BATCH_NO_AUTOSTATE(6); OUT_BATCH_REGSEQ(R300_ZB_DEPTHOFFSET, 1); OUT_BATCH_RELOC(0, rrbd->bo, 0, 0, RADEON_GEM_DOMAIN_VRAM, 0); - OUT_BATCH_REGVAL(R300_ZB_DEPTHPITCH, cbpitch); + OUT_BATCH_REGSEQ(R300_ZB_DEPTHPITCH, 1); + if (!r300->radeon.radeonScreen->kernel_mm) + OUT_BATCH(zbpitch); + else + OUT_BATCH_RELOC(zbpitch, rrbd->bo, zbpitch, 0, RADEON_GEM_DOMAIN_VRAM, 0); END_BATCH(); } #endif diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c index 1ac14267d5e..050e8cd2a7d 100644 --- a/src/mesa/drivers/dri/r300/r300_state.c +++ b/src/mesa/drivers/dri/r300/r300_state.c @@ -1253,6 +1253,7 @@ static GLuint translate_lod_bias(GLfloat bias) return (((GLuint)b) << R300_LOD_BIAS_SHIFT) & R300_LOD_BIAS_MASK; } + static void r300SetupTextures(GLcontext * ctx) { int i, mtu; @@ -1345,6 +1346,28 @@ static void r300SetupTextures(GLcontext * ctx) } } + /* R3xx and R4xx chips require that the texture unit corresponding to + * KIL instructions is really enabled. + * + * We do some fakery here and in the state atom emit logic to enable + * the texture without tripping up the CS checker in the kernel. + */ + if (r300->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV515) { + if (ctx->FragmentProgram._Current->UsesKill && last_hw_tmu < 0) { + last_hw_tmu++; + + r300->hw.txe.cmd[R300_TXE_ENABLE] |= 1; + + r300->hw.tex.border_color.cmd[R300_TEX_VALUE_0] = 0; + r300->hw.tex.chroma_key.cmd[R300_TEX_VALUE_0] = 0; + r300->hw.tex.filter.cmd[R300_TEX_VALUE_0] = 0; + r300->hw.tex.filter_1.cmd[R300_TEX_VALUE_0] = 0; + r300->hw.tex.size.cmd[R300_TEX_VALUE_0] = 0; /* 1x1 texture */ + r300->hw.tex.format.cmd[R300_TEX_VALUE_0] = 0; /* A8 format */ + r300->hw.tex.pitch.cmd[R300_TEX_VALUE_0] = 0; + } + } + r300->hw.tex.filter.cmd[R300_TEX_CMD_0] = cmdpacket0(r300->radeon.radeonScreen, R300_TX_FILTER0_0, last_hw_tmu + 1); r300->hw.tex.filter_1.cmd[R300_TEX_CMD_0] = @@ -1362,16 +1385,6 @@ static void r300SetupTextures(GLcontext * ctx) r300->hw.tex.border_color.cmd[R300_TEX_CMD_0] = cmdpacket0(r300->radeon.radeonScreen, R300_TX_BORDER_COLOR_0, last_hw_tmu + 1); - if (r300->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV515) { - if (ctx->FragmentProgram._Current->UsesKill && last_hw_tmu < 0) { - // The KILL operation requires the first texture unit - // to be enabled. - r300->hw.txe.cmd[R300_TXE_ENABLE] |= 1; - r300->hw.tex.filter.cmd[R300_TEX_VALUE_0] = 0; - r300->hw.tex.filter.cmd[R300_TEX_CMD_0] = - cmdpacket0(r300->radeon.radeonScreen, R300_TX_FILTER0_0, 1); - } - } r300->vtbl.SetupFragmentShaderTextures(ctx, tmu_mappings); if (RADEON_DEBUG & DEBUG_STATE) @@ -2103,13 +2116,13 @@ static void r500SetupPixelShader(GLcontext *ctx) rmesa->hw.fp.cmd[R500_FP_PIXSIZE] = code->max_temp_idx; rmesa->hw.fp.cmd[R500_FP_CODE_ADDR] = - R500_US_CODE_START_ADDR(code->inst_offset) | + R500_US_CODE_START_ADDR(0) | R500_US_CODE_END_ADDR(code->inst_end); rmesa->hw.fp.cmd[R500_FP_CODE_RANGE] = - R500_US_CODE_RANGE_ADDR(code->inst_offset) | + R500_US_CODE_RANGE_ADDR(0) | R500_US_CODE_RANGE_SIZE(code->inst_end); rmesa->hw.fp.cmd[R500_FP_CODE_OFFSET] = - R500_US_CODE_OFFSET_ADDR(0); /* FIXME when we add flow control */ + R500_US_CODE_OFFSET_ADDR(0); R300_STATECHANGE(rmesa, r500fp); /* Emit our shader... */ @@ -2234,6 +2247,7 @@ static void r300InvalidateState(GLcontext * ctx, GLuint new_state) _mesa_update_draw_buffer_bounds(ctx); R300_STATECHANGE(r300, cb); + R300_STATECHANGE(r300, zb); } r300->radeon.NewGLState |= new_state; diff --git a/src/mesa/drivers/dri/r600/r600_cmdbuf.c b/src/mesa/drivers/dri/r600/r600_cmdbuf.c index 74fec02584b..dc2fb0144a2 100644 --- a/src/mesa/drivers/dri/r600/r600_cmdbuf.c +++ b/src/mesa/drivers/dri/r600/r600_cmdbuf.c @@ -78,8 +78,7 @@ int r600_cs_write_reloc(struct radeon_cs *cs, struct radeon_bo *bo, uint32_t read_domain, uint32_t write_domain, - uint32_t flags, - offset_modifiers* poffset_mod) + uint32_t flags) { struct r600_cs_reloc_legacy *relocs; int i; @@ -135,10 +134,6 @@ int r600_cs_write_reloc(struct radeon_cs *cs, cs->section_ndw += 2; cs->section_cdw += 2; - relocs[i].offset_mod.shift = poffset_mod->shift; - relocs[i].offset_mod.shiftbits = poffset_mod->shiftbits; - relocs[i].offset_mod.mask = poffset_mod->mask; - return 0; } } @@ -160,9 +155,6 @@ int r600_cs_write_reloc(struct radeon_cs *cs, { return -ENOMEM; } - relocs[cs->crelocs].offset_mod.shift = poffset_mod->shift; - relocs[cs->crelocs].offset_mod.shiftbits = poffset_mod->shiftbits; - relocs[cs->crelocs].offset_mod.mask = poffset_mod->mask; relocs[cs->crelocs].indices[0] = cs->cdw - 1; relocs[cs->crelocs].reloc_indices[0] = cs->section_cdw; @@ -255,65 +247,44 @@ static int r600_cs_process_relocs(struct radeon_cs *cs, csm = (struct r600_cs_manager_legacy*)cs->csm; relocs = (struct r600_cs_reloc_legacy *)cs->relocs; restart: - for (i = 0; i < cs->crelocs; i++) - { - for (j = 0; j < relocs[i].cindices; j++) - { + for (i = 0; i < cs->crelocs; i++) { uint32_t soffset, eoffset, asicoffset; r = radeon_bo_legacy_validate(relocs[i].base.bo, - &soffset, &eoffset); - if (r == -EAGAIN) - { - goto restart; + &soffset, &eoffset); + if (r == -EAGAIN) { + goto restart; } - if (r) - { - fprintf(stderr, "validated %p [0x%08X, 0x%08X]\n", - relocs[i].base.bo, soffset, eoffset); - return r; + if (r) { + fprintf(stderr, "validated %p [0x%08X, 0x%08X]\n", + relocs[i].base.bo, soffset, eoffset); + return r; } asicoffset = soffset; - if (asicoffset >= eoffset) - { - /* radeon_bo_debug(relocs[i].base.bo, 12); */ - fprintf(stderr, "validated %p [0x%08X, 0x%08X]\n", - relocs[i].base.bo, soffset, eoffset); - fprintf(stderr, "above end: %p 0x%08X 0x%08X\n", - relocs[i].base.bo, - cs->packets[relocs[i].indices[j]], - eoffset); - exit(0); - return -EINVAL; - } - /* apply offset operator */ - switch (relocs[i].offset_mod.shift) - { - case NO_SHIFT: - asicoffset = asicoffset & relocs[i].offset_mod.mask; - break; - case LEFT_SHIFT: - asicoffset = (asicoffset << relocs[i].offset_mod.shiftbits) & relocs[i].offset_mod.mask; - break; - case RIGHT_SHIFT: - asicoffset = (asicoffset >> relocs[i].offset_mod.shiftbits) & relocs[i].offset_mod.mask; - break; - default: - break; - }; - - /* pkt3 nop header in ib chunk */ - cs->packets[relocs[i].reloc_indices[j]] = 0xC0001000; - - /* reloc index in ib chunk */ - cs->packets[relocs[i].reloc_indices[j] + 1] = offset_dw; - - /* asic offset in reloc chunk */ /* see alex drm r600_nomm_relocate */ - reloc_chunk[offset_dw] = asicoffset; - reloc_chunk[offset_dw + 3] = 0; - - offset_dw += 4; - } + + for (j = 0; j < relocs[i].cindices; j++) { + if (asicoffset >= eoffset) { + /* radeon_bo_debug(relocs[i].base.bo, 12); */ + fprintf(stderr, "validated %p [0x%08X, 0x%08X]\n", + relocs[i].base.bo, soffset, eoffset); + fprintf(stderr, "above end: %p 0x%08X 0x%08X\n", + relocs[i].base.bo, + cs->packets[relocs[i].indices[j]], + eoffset); + exit(0); + return -EINVAL; + } + /* pkt3 nop header in ib chunk */ + cs->packets[relocs[i].reloc_indices[j]] = 0xC0001000; + /* reloc index in ib chunk */ + cs->packets[relocs[i].reloc_indices[j] + 1] = offset_dw; + } + + /* asic offset in reloc chunk */ /* see alex drm r600_nomm_relocate */ + reloc_chunk[offset_dw] = asicoffset; + reloc_chunk[offset_dw + 3] = 0; + + offset_dw += 4; } *length_dw_reloc_chunk = offset_dw; @@ -351,10 +322,7 @@ static int r600_cs_emit(struct radeon_cs *cs) struct r600_cs_manager_legacy *csm = (struct r600_cs_manager_legacy*)cs->csm; struct drm_radeon_cs cs_cmd; struct drm_radeon_cs_chunk cs_chunk[2]; - drm_radeon_cmd_buffer_t cmd; - /* drm_r300_cmd_header_t age; */ uint32_t length_dw_reloc_chunk; - uint64_t ull; uint64_t chunk_ptrs[2]; uint32_t reloc_chunk[128]; int r; @@ -363,43 +331,13 @@ static int r600_cs_emit(struct radeon_cs *cs) /* TODO : put chip level things here if need. */ /* csm->ctx->vtbl.emit_cs_header(cs, csm->ctx); */ - BATCH_LOCALS(csm->ctx); - drm_radeon_getparam_t gp; - uint32_t current_scratchx_age; - - gp.param = RADEON_PARAM_LAST_CLEAR; - gp.value = (int *)¤t_scratchx_age; - r = drmCommandWriteRead(cs->csm->fd, - DRM_RADEON_GETPARAM, - &gp, - sizeof(gp)); - if (r) - { - fprintf(stderr, "%s: drmRadeonGetParam: %d\n", __FUNCTION__, r); - exit(1); - } - - csm->pending_age = 0; csm->pending_count = 1; - current_scratchx_age++; - csm->pending_age = current_scratchx_age; - - BEGIN_BATCH_NO_AUTOSTATE(3); - R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); - R600_OUT_BATCH((SCRATCH_REG2 - R600_SET_CONFIG_REG_OFFSET) >> 2); - R600_OUT_BATCH(current_scratchx_age); - END_BATCH(); - COMMIT_BATCH(); - - //TODO ioctl to get back cs id assigned in drm - //csm->pending_age = cs_id_back; - r = r600_cs_process_relocs(cs, &(reloc_chunk[0]), &length_dw_reloc_chunk); if (r) { return 0; } - + /* raw ib chunk */ cs_chunk[0].chunk_id = RADEON_CHUNK_ID_IB; cs_chunk[0].length_dw = cs->cdw; @@ -429,6 +367,8 @@ static int r600_cs_emit(struct radeon_cs *cs) return r; } + csm->pending_age = cs_cmd.cs_id; + r600_cs_set_age(cs); cs->csm->read_used = 0; @@ -514,8 +454,12 @@ struct radeon_cs_manager * r600_radeon_cs_manager_legacy_ctor(struct radeon_cont void r600InitCmdBuf(context_t *r600) /* from rcommonInitCmdBuf */ { radeonContextPtr rmesa = &r600->radeon; - GLuint size; + rmesa->hw.max_state_size = 4000; /* rough estimate */ + + rmesa->hw.all_dirty = GL_TRUE; + rmesa->hw.is_dirty = GL_TRUE; + /* Initialize command buffer */ size = 256 * driQueryOptioni(&rmesa->optionCache, "command_buffer_size"); diff --git a/src/mesa/drivers/dri/r600/r600_cmdbuf.h b/src/mesa/drivers/dri/r600/r600_cmdbuf.h index bd1ed7fdff0..5df0cf1ab61 100644 --- a/src/mesa/drivers/dri/r600/r600_cmdbuf.h +++ b/src/mesa/drivers/dri/r600/r600_cmdbuf.h @@ -132,15 +132,13 @@ struct r600_cs_reloc_legacy { uint32_t cindices; uint32_t *indices; uint32_t *reloc_indices; - struct offset_modifiers offset_mod; }; extern int r600_cs_write_reloc(struct radeon_cs *cs, struct radeon_bo *bo, uint32_t read_domain, uint32_t write_domain, - uint32_t flags, - offset_modifiers* poffset_mod); + uint32_t flags); static inline void r600_cs_write_dword(struct radeon_cs *cs, uint32_t dword) { @@ -171,7 +169,7 @@ struct radeon_cs_manager * r600_radeon_cs_manager_legacy_ctor(struct radeon_cont /** * Write a relocated dword to the command buffer. */ -#define R600_OUT_BATCH_RELOC(data, bo, offset, rd, wd, flags, offset_mod) \ +#define R600_OUT_BATCH_RELOC(data, bo, offset, rd, wd, flags) \ do { \ if (0 && offset) { \ fprintf(stderr, "(%s:%s:%d) offset : %d\n", \ @@ -179,7 +177,7 @@ struct radeon_cs_manager * r600_radeon_cs_manager_legacy_ctor(struct radeon_cont } \ r600_cs_write_dword(b_l_rmesa->cmdbuf.cs, offset); \ r600_cs_write_reloc(b_l_rmesa->cmdbuf.cs, \ - bo, rd, wd, flags, offset_mod); \ + bo, rd, wd, flags); \ } while(0) /* R600/R700 */ diff --git a/src/mesa/drivers/dri/r600/r600_context.c b/src/mesa/drivers/dri/r600/r600_context.c index eacf8112825..7009374b0ca 100644 --- a/src/mesa/drivers/dri/r600/r600_context.c +++ b/src/mesa/drivers/dri/r600/r600_context.c @@ -395,6 +395,8 @@ r600DestroyContext (__DRIcontextPrivate * driContextPriv) if (context) FREE(context->hw.pStateList); + + radeonDestroyContext(driContextPriv); } diff --git a/src/mesa/drivers/dri/r600/r600_context.h b/src/mesa/drivers/dri/r600/r600_context.h index bcb33e1386f..30ddce682c1 100644 --- a/src/mesa/drivers/dri/r600/r600_context.h +++ b/src/mesa/drivers/dri/r600/r600_context.h @@ -128,13 +128,6 @@ enum RIGHT_SHIFT = 2, }; -typedef struct offset_modifiers -{ - GLuint shift; - GLuint shiftbits; - GLuint mask; -} offset_modifiers; - /** * \brief R600 context structure. */ @@ -165,6 +158,17 @@ extern GLboolean r700InitChipObject(context_t *context); extern GLboolean r700SendContextStates(context_t *context); extern GLboolean r700SendViewportState(context_t *context, int id); extern GLboolean r700SendRenderTargetState(context_t *context, int id); +extern GLboolean r700SendTextureState(context_t *context); +extern GLboolean r700SendDepthTargetState(context_t *context); +extern GLboolean r700SendUCPState(context_t *context); +extern GLboolean r700SendFSState(context_t *context); +extern void r700EmitState(GLcontext * ctx); + +extern GLboolean r700SyncSurf(context_t *context, + struct radeon_bo *pbo, + uint32_t read_domain, + uint32_t write_domain, + uint32_t sync_type); extern int r700SetupStreams(GLcontext * ctx); extern void r700SetupVTXConstants(GLcontext * ctx, diff --git a/src/mesa/drivers/dri/r600/r600_tex.c b/src/mesa/drivers/dri/r600/r600_tex.c index 853f824b2c0..6d531bf0f93 100644 --- a/src/mesa/drivers/dri/r600/r600_tex.c +++ b/src/mesa/drivers/dri/r600/r600_tex.c @@ -160,7 +160,7 @@ static void r600SetTexDefaultState(radeonTexObjPtr t) SETfield(t->SQ_TEX_SAMPLER0, SQ_TEX_BORDER_COLOR_TRANS_BLACK, BORDER_COLOR_TYPE_shift, BORDER_COLOR_TYPE_mask); t->SQ_TEX_SAMPLER1 = 0; - SETfield(t->SQ_TEX_SAMPLER1, MAX_LOD_mask, MAX_LOD_shift, MAX_LOD_mask); + SETfield(t->SQ_TEX_SAMPLER1, 0x3ff, MAX_LOD_shift, MAX_LOD_mask); t->SQ_TEX_SAMPLER2 = 0; SETbit(t->SQ_TEX_SAMPLER2, SQ_TEX_SAMPLER_WORD2_0__TYPE_bit); diff --git a/src/mesa/drivers/dri/r600/r600_texstate.c b/src/mesa/drivers/dri/r600/r600_texstate.c index 1cf3b484ae6..ee9b64ee43a 100644 --- a/src/mesa/drivers/dri/r600/r600_texstate.c +++ b/src/mesa/drivers/dri/r600/r600_texstate.c @@ -556,7 +556,7 @@ static void setup_hardware_state(context_t *rmesa, struct gl_texture_object *tex radeonTexObj *t = radeon_tex_obj(texObj); const struct gl_texture_image *firstImage; int firstlevel = t->mt ? t->mt->firstLevel : 0; - GLuint uTexelPitch; + GLuint uTexelPitch, row_align; firstImage = t->base.Image[0][firstlevel]; @@ -568,9 +568,6 @@ static void setup_hardware_state(context_t *rmesa, struct gl_texture_object *tex } } - if (t->image_override && t->bo) - return; - switch (texObj->Target) { case GL_TEXTURE_1D: SETfield(t->SQ_TEX_RESOURCE0, SQ_TEX_DIM_1D, DIM_shift, DIM_mask); @@ -595,15 +592,26 @@ static void setup_hardware_state(context_t *rmesa, struct gl_texture_object *tex return; } - uTexelPitch = (firstImage->Width + R700_TEXEL_PITCH_ALIGNMENT_MASK) + row_align = rmesa->radeon.texture_row_align - 1; + uTexelPitch = ((firstImage->Width * t->mt->bpp + row_align) & ~row_align) / t->mt->bpp; + uTexelPitch = (uTexelPitch + R700_TEXEL_PITCH_ALIGNMENT_MASK) & ~R700_TEXEL_PITCH_ALIGNMENT_MASK; + /* min pitch is 8 */ + if (uTexelPitch < 8) + uTexelPitch = 8; + SETfield(t->SQ_TEX_RESOURCE0, (uTexelPitch/8)-1, PITCH_shift, PITCH_mask); SETfield(t->SQ_TEX_RESOURCE0, firstImage->Width - 1, TEX_WIDTH_shift, TEX_WIDTH_mask); SETfield(t->SQ_TEX_RESOURCE1, firstImage->Height - 1, TEX_HEIGHT_shift, TEX_HEIGHT_mask); + if ((t->mt->lastLevel - t->mt->firstLevel) > 0) { + t->SQ_TEX_RESOURCE3 = t->mt->levels[0].size / 256; + SETfield(t->SQ_TEX_RESOURCE4, t->mt->firstLevel, BASE_LEVEL_shift, BASE_LEVEL_mask); + SETfield(t->SQ_TEX_RESOURCE5, t->mt->lastLevel, LAST_LEVEL_shift, LAST_LEVEL_mask); + } } /** @@ -690,7 +698,7 @@ void r600SetTexOffset(__DRIcontext * pDRICtx, GLint texname, struct gl_texture_object *tObj = _mesa_lookup_texture(rmesa->radeon.glCtx, texname); radeonTexObjPtr t = radeon_tex_obj(tObj); - uint32_t pitch_val; + uint32_t pitch_val, size; if (!tObj) return; @@ -700,7 +708,12 @@ void r600SetTexOffset(__DRIcontext * pDRICtx, GLint texname, if (!offset) return; - t->bo = NULL; + size = pitch;//h * w * (depth / 8); + if (t->bo) { + radeon_bo_unref(t->bo); + t->bo = NULL; + } + t->bo = radeon_legacy_bo_alloc_fake(rmesa->radeon.radeonScreen->bom, size, offset); t->override_offset = offset; pitch_val = pitch; switch (depth) { @@ -751,6 +764,11 @@ void r600SetTexOffset(__DRIcontext * pDRICtx, GLint texname, pitch_val = (pitch_val + R700_TEXEL_PITCH_ALIGNMENT_MASK) & ~R700_TEXEL_PITCH_ALIGNMENT_MASK; + + /* min pitch is 8 */ + if (pitch_val < 8) + pitch_val = 8; + SETfield(t->SQ_TEX_RESOURCE0, (pitch_val/8)-1, PITCH_shift, PITCH_mask); } @@ -898,6 +916,10 @@ void r600SetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_fo pitch_val = (pitch_val + R700_TEXEL_PITCH_ALIGNMENT_MASK) & ~R700_TEXEL_PITCH_ALIGNMENT_MASK; + /* min pitch is 8 */ + if (pitch_val < 8) + pitch_val = 8; + SETfield(t->SQ_TEX_RESOURCE0, (pitch_val/8)-1, PITCH_shift, PITCH_mask); SETfield(t->SQ_TEX_RESOURCE0, rb->base.Width - 1, TEX_WIDTH_shift, TEX_WIDTH_mask); diff --git a/src/mesa/drivers/dri/r600/r700_assembler.c b/src/mesa/drivers/dri/r600/r700_assembler.c index ebd5ff106be..eaacd061137 100644 --- a/src/mesa/drivers/dri/r600/r700_assembler.c +++ b/src/mesa/drivers/dri/r600/r700_assembler.c @@ -1205,7 +1205,7 @@ GLboolean tex_src(r700_AssemblerBase *pAsm) return GL_TRUE; } -GLboolean assemble_tex_instruction(r700_AssemblerBase *pAsm) +GLboolean assemble_tex_instruction(r700_AssemblerBase *pAsm, GLboolean normalized) { PVSSRC * texture_coordinate_source; PVSSRC * texture_unit_source; @@ -1227,10 +1227,18 @@ GLboolean assemble_tex_instruction(r700_AssemblerBase *pAsm) tex_instruction_ptr->m_Word0.f.resource_id = texture_unit_source->reg; tex_instruction_ptr->m_Word1.f.lod_bias = 0x0; - tex_instruction_ptr->m_Word1.f.coord_type_x = SQ_TEX_NORMALIZED; - tex_instruction_ptr->m_Word1.f.coord_type_y = SQ_TEX_NORMALIZED; - tex_instruction_ptr->m_Word1.f.coord_type_z = SQ_TEX_NORMALIZED; - tex_instruction_ptr->m_Word1.f.coord_type_w = SQ_TEX_NORMALIZED; + if (normalized) { + tex_instruction_ptr->m_Word1.f.coord_type_x = SQ_TEX_NORMALIZED; + tex_instruction_ptr->m_Word1.f.coord_type_y = SQ_TEX_NORMALIZED; + tex_instruction_ptr->m_Word1.f.coord_type_z = SQ_TEX_NORMALIZED; + tex_instruction_ptr->m_Word1.f.coord_type_w = SQ_TEX_NORMALIZED; + } else { + /* XXX: UNNORMALIZED tex coords have limited wrap modes */ + tex_instruction_ptr->m_Word1.f.coord_type_x = SQ_TEX_UNNORMALIZED; + tex_instruction_ptr->m_Word1.f.coord_type_y = SQ_TEX_UNNORMALIZED; + tex_instruction_ptr->m_Word1.f.coord_type_z = SQ_TEX_UNNORMALIZED; + tex_instruction_ptr->m_Word1.f.coord_type_w = SQ_TEX_UNNORMALIZED; + } tex_instruction_ptr->m_Word2.f.offset_x = 0x0; tex_instruction_ptr->m_Word2.f.offset_y = 0x0; @@ -2196,11 +2204,19 @@ GLboolean next_ins(r700_AssemblerBase *pAsm) if( GL_TRUE == IsTex(pILInst->Opcode) ) { - if( GL_FALSE == assemble_tex_instruction(pAsm) ) - { - r700_error(ERROR_ASM_TEXINSTRUCTION, "Error assembling TEX instruction"); - return GL_FALSE; - } + if (pILInst->TexSrcTarget == TEXTURE_RECT_INDEX) { + if( GL_FALSE == assemble_tex_instruction(pAsm, GL_FALSE) ) + { + r700_error(ERROR_ASM_TEXINSTRUCTION, "Error assembling TEX instruction"); + return GL_FALSE; + } + } else { + if( GL_FALSE == assemble_tex_instruction(pAsm, GL_TRUE) ) + { + r700_error(ERROR_ASM_TEXINSTRUCTION, "Error assembling TEX instruction"); + return GL_FALSE; + } + } } else { //ALU @@ -3823,6 +3839,9 @@ GLboolean Process_Export(r700_AssemblerBase* pAsm, if (export_count == 1) { ucWriteMask = pAsm->pucOutMask[starting_register_number - pAsm->starting_export_register_number]; + /* exports Z as a float into Red channel */ + if (GL_TRUE == is_depth_export) + ucWriteMask = 0x1; if( (ucWriteMask & 0x1) != 0) { diff --git a/src/mesa/drivers/dri/r600/r700_assembler.h b/src/mesa/drivers/dri/r600/r700_assembler.h index e9b21b802ec..f9c4d849c65 100644 --- a/src/mesa/drivers/dri/r600/r700_assembler.h +++ b/src/mesa/drivers/dri/r600/r700_assembler.h @@ -424,7 +424,7 @@ GLboolean assemble_src(r700_AssemblerBase *pAsm, GLboolean assemble_dst(r700_AssemblerBase *pAsm); GLboolean tex_dst(r700_AssemblerBase *pAsm); GLboolean tex_src(r700_AssemblerBase *pAsm); -GLboolean assemble_tex_instruction(r700_AssemblerBase *pAsm); +GLboolean assemble_tex_instruction(r700_AssemblerBase *pAsm, GLboolean normalized); void initialize(r700_AssemblerBase *pAsm); GLboolean assemble_alu_src(R700ALUInstruction* alu_instruction_ptr, int source_index, diff --git a/src/mesa/drivers/dri/r600/r700_chip.c b/src/mesa/drivers/dri/r600/r700_chip.c index c083862f369..e5e0f556cbe 100644 --- a/src/mesa/drivers/dri/r600/r700_chip.c +++ b/src/mesa/drivers/dri/r600/r700_chip.c @@ -38,6 +38,8 @@ #include "r700_vertprog.h" #include "r700_ioctl.h" +#include "radeon_mipmap_tree.h" + #define LINK_STATES(reg) \ do \ { \ @@ -241,6 +243,67 @@ GLboolean r700InitChipObject(context_t *context) return GL_TRUE; } +GLboolean r700SendTextureState(context_t *context) +{ + unsigned int i; + R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw); + struct radeon_bo *bo = NULL; + BATCH_LOCALS(&context->radeon); + + for (i=0; i<R700_TEXTURE_NUMBERUNITS; i++) { + radeonTexObj *t = r700->textures[i]; + if (t) { + if (!t->image_override) + bo = t->mt->bo; + else + bo = t->bo; + if (bo) { + + r700SyncSurf(context, bo, + RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, + 0, TC_ACTION_ENA_bit); + + BEGIN_BATCH_NO_AUTOSTATE(9); + R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7)); + R600_OUT_BATCH(i * 7); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE0); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE1); + R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE2, + bo, + 0, + RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0); + R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE3, + bo, + r700->textures[i]->SQ_TEX_RESOURCE3, + RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE4); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE5); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE6); + END_BATCH(); + + BEGIN_BATCH_NO_AUTOSTATE(5); + R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_SAMPLER, 3)); + R600_OUT_BATCH(i * 3); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER0); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER1); + R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER2); + END_BATCH(); + + BEGIN_BATCH_NO_AUTOSTATE(2 + 4); + R600_OUT_BATCH_REGSEQ((TD_PS_SAMPLER0_BORDER_RED + (i * 16)), 4); + R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_RED); + R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_GREEN); + R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_BLUE); + R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_ALPHA); + END_BATCH(); + + COMMIT_BATCH(); + } + } + } + return GL_TRUE; +} + void r700SetupVTXConstants(GLcontext * ctx, unsigned int nStreamID, void * pAos, @@ -249,10 +312,7 @@ void r700SetupVTXConstants(GLcontext * ctx, unsigned int count) /* number of vectors in stream */ { context_t *context = R700_CONTEXT(ctx); - uint32_t *dest; struct radeon_aos * paos = (struct radeon_aos *)pAos; - offset_modifiers offset_mod = {NO_SHIFT, 0, 0xFFFFFFFF}; - BATCH_LOCALS(&context->radeon); unsigned int uSQ_VTX_CONSTANT_WORD0_0; @@ -261,6 +321,9 @@ void r700SetupVTXConstants(GLcontext * ctx, unsigned int uSQ_VTX_CONSTANT_WORD3_0 = 0; unsigned int uSQ_VTX_CONSTANT_WORD6_0 = 0; + if (!paos->bo) + return; + if ((context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV610) || (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV620) || (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RS780) || @@ -294,7 +357,7 @@ void r700SetupVTXConstants(GLcontext * ctx, R600_OUT_BATCH_RELOC(uSQ_VTX_CONSTANT_WORD0_0, paos->bo, uSQ_VTX_CONSTANT_WORD0_0, - RADEON_GEM_DOMAIN_GTT, 0, 0, &offset_mod); + RADEON_GEM_DOMAIN_GTT, 0, 0); R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD1_0); R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD2_0); R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD3_0); @@ -310,7 +373,6 @@ void r700SetupVTXConstants(GLcontext * ctx, int r700SetupStreams(GLcontext * ctx) { context_t *context = R700_CONTEXT(ctx); - BATCH_LOCALS(&context->radeon); struct r700_vertex_program *vpc @@ -324,7 +386,7 @@ int r700SetupStreams(GLcontext * ctx) BEGIN_BATCH_NO_AUTOSTATE(6); R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_CTL_CONST, 1)); - R600_OUT_BATCH(mmSQ_VTX_BASE_VTX_LOC - ASIC_CTL_CONST_BASE_INDEX); + R600_OUT_BATCH(mmSQ_VTX_BASE_VTX_LOC - ASIC_CTL_CONST_BASE_INDEX); R600_OUT_BATCH(0); R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_CTL_CONST, 1)); @@ -417,11 +479,10 @@ GLboolean r700SendContextStates(context_t *context) return GL_TRUE; } -GLboolean r700SendDepthTargetState(context_t *context, int id) +GLboolean r700SendDepthTargetState(context_t *context) { R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context); struct radeon_renderbuffer *rrb; - offset_modifiers offset_mod; BATCH_LOCALS(&context->radeon); rrb = radeon_get_depthbuffer(&context->radeon); @@ -430,10 +491,6 @@ GLboolean r700SendDepthTargetState(context_t *context, int id) return GL_FALSE; } - offset_mod.shift = NO_SHIFT; - offset_mod.shiftbits = 0; - offset_mod.mask = 0xFFFFFFFF; - BEGIN_BATCH_NO_AUTOSTATE(9); R600_OUT_BATCH_REGSEQ(DB_DEPTH_SIZE, 2); R600_OUT_BATCH(r700->DB_DEPTH_SIZE.u32All); @@ -442,7 +499,7 @@ GLboolean r700SendDepthTargetState(context_t *context, int id) R600_OUT_BATCH_RELOC(r700->DB_DEPTH_BASE.u32All, rrb->bo, r700->DB_DEPTH_BASE.u32All, - 0, RADEON_GEM_DOMAIN_VRAM, 0, &offset_mod); + 0, RADEON_GEM_DOMAIN_VRAM, 0); R600_OUT_BATCH(r700->DB_DEPTH_INFO.u32All); R600_OUT_BATCH(r700->DB_HTILE_DATA_BASE.u32All); END_BATCH(); @@ -479,7 +536,6 @@ GLboolean r700SendRenderTargetState(context_t *context, int id) { R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context); struct radeon_renderbuffer *rrb; - offset_modifiers offset_mod; BATCH_LOCALS(&context->radeon); rrb = radeon_get_colorbuffer(&context->radeon); @@ -494,16 +550,12 @@ GLboolean r700SendRenderTargetState(context_t *context, int id) if (!r700->render_target[id].enabled) return GL_FALSE; - offset_mod.shift = NO_SHIFT; - offset_mod.shiftbits = 0; - offset_mod.mask = 0xFFFFFFFF; - BEGIN_BATCH_NO_AUTOSTATE(3); R600_OUT_BATCH_REGSEQ(CB_COLOR0_BASE + (4 * id), 1); R600_OUT_BATCH_RELOC(r700->render_target[id].CB_COLOR0_BASE.u32All, rrb->bo, r700->render_target[id].CB_COLOR0_BASE.u32All, - 0, RADEON_GEM_DOMAIN_VRAM, 0, &offset_mod); + 0, RADEON_GEM_DOMAIN_VRAM, 0); END_BATCH(); if ((context->radeon.radeonScreen->chip_family > CHIP_FAMILY_R600) && @@ -540,16 +592,13 @@ GLboolean r700SendRenderTargetState(context_t *context, int id) GLboolean r700SendPSState(context_t *context) { R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context); - struct radeon_renderbuffer *rrb; struct radeon_bo * pbo; - offset_modifiers offset_mod; BATCH_LOCALS(&context->radeon); pbo = (struct radeon_bo *)r700GetActiveFpShaderBo(GL_CONTEXT(context)); - offset_mod.shift = NO_SHIFT; - offset_mod.shiftbits = 0; - offset_mod.mask = 0xFFFFFFFF; + if (!pbo) + return GL_FALSE; r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit); @@ -558,7 +607,7 @@ GLboolean r700SendPSState(context_t *context) R600_OUT_BATCH_RELOC(r700->ps.SQ_PGM_START_PS.u32All, pbo, r700->ps.SQ_PGM_START_PS.u32All, - RADEON_GEM_DOMAIN_GTT, 0, 0, &offset_mod); + RADEON_GEM_DOMAIN_GTT, 0, 0); END_BATCH(); BEGIN_BATCH_NO_AUTOSTATE(9); @@ -575,16 +624,13 @@ GLboolean r700SendPSState(context_t *context) GLboolean r700SendVSState(context_t *context) { R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context); - struct radeon_renderbuffer *rrb; struct radeon_bo * pbo; - offset_modifiers offset_mod; BATCH_LOCALS(&context->radeon); pbo = (struct radeon_bo *)r700GetActiveVpShaderBo(GL_CONTEXT(context)); - offset_mod.shift = NO_SHIFT; - offset_mod.shiftbits = 0; - offset_mod.mask = 0xFFFFFFFF; + if (!pbo) + return GL_FALSE; r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit); @@ -593,7 +639,7 @@ GLboolean r700SendVSState(context_t *context) R600_OUT_BATCH_RELOC(r700->vs.SQ_PGM_START_VS.u32All, pbo, r700->vs.SQ_PGM_START_VS.u32All, - RADEON_GEM_DOMAIN_GTT, 0, 0, &offset_mod); + RADEON_GEM_DOMAIN_GTT, 0, 0); END_BATCH(); BEGIN_BATCH_NO_AUTOSTATE(6); @@ -609,9 +655,7 @@ GLboolean r700SendVSState(context_t *context) GLboolean r700SendFSState(context_t *context) { R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context); - struct radeon_renderbuffer *rrb; struct radeon_bo * pbo; - offset_modifiers offset_mod; BATCH_LOCALS(&context->radeon); /* XXX fixme @@ -625,9 +669,8 @@ GLboolean r700SendFSState(context_t *context) r700->fs.SQ_PGM_CF_OFFSET_FS.u32All = 0; /* XXX */ - offset_mod.shift = NO_SHIFT; - offset_mod.shiftbits = 0; - offset_mod.mask = 0xFFFFFFFF; + if (!pbo) + return GL_FALSE; r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit); @@ -636,7 +679,7 @@ GLboolean r700SendFSState(context_t *context) R600_OUT_BATCH_RELOC(r700->fs.SQ_PGM_START_FS.u32All, pbo, r700->fs.SQ_PGM_START_FS.u32All, - RADEON_GEM_DOMAIN_GTT, 0, 0, &offset_mod); + RADEON_GEM_DOMAIN_GTT, 0, 0); END_BATCH(); BEGIN_BATCH_NO_AUTOSTATE(6); @@ -652,8 +695,6 @@ GLboolean r700SendFSState(context_t *context) GLboolean r700SendViewportState(context_t *context, int id) { R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context); - struct radeon_renderbuffer *rrb; - offset_modifiers offset_mod; BATCH_LOCALS(&context->radeon); if (id > R700_MAX_VIEWPORTS) diff --git a/src/mesa/drivers/dri/r600/r700_clear.c b/src/mesa/drivers/dri/r600/r700_clear.c index e84be386221..05d4af331e0 100644 --- a/src/mesa/drivers/dri/r600/r700_clear.c +++ b/src/mesa/drivers/dri/r600/r700_clear.c @@ -31,6 +31,7 @@ #include "main/imports.h" #include "main/mtypes.h" #include "main/enums.h" +#include "swrast/swrast.h" #include "radeon_lock.h" #include "r600_context.h" diff --git a/src/mesa/drivers/dri/r600/r700_fragprog.c b/src/mesa/drivers/dri/r600/r700_fragprog.c index 180d980442b..6249bde6f18 100644 --- a/src/mesa/drivers/dri/r600/r700_fragprog.c +++ b/src/mesa/drivers/dri/r600/r700_fragprog.c @@ -55,6 +55,12 @@ void Map_Fragment_Program(r700_AssemblerBase *pAsm, //Input mapping : mesa_fp->Base.InputsRead set the flag, set in //The flags parsed in parse_attrib_binding. FRAG_ATTRIB_COLx, FRAG_ATTRIB_TEXx, ... //MUST match order in Map_Vertex_Output + unBit = 1 << FRAG_ATTRIB_WPOS; + if(mesa_fp->Base.InputsRead & unBit) + { + pAsm->uiFP_AttributeMap[FRAG_ATTRIB_WPOS] = pAsm->number_used_registers++; + } + unBit = 1 << FRAG_ATTRIB_COL0; if(mesa_fp->Base.InputsRead & unBit) { @@ -112,6 +118,7 @@ void Map_Fragment_Program(r700_AssemblerBase *pAsm, pAsm->uiFP_OutputMap[FRAG_RESULT_DEPTH] = pAsm->number_used_registers++; pAsm->number_of_exports++; pAsm->number_of_colorandz_exports++; + pAsm->pR700Shader->depthIsExported = 1; } pAsm->pucOutMask = (unsigned char*) MALLOC(pAsm->number_of_exports); @@ -262,9 +269,8 @@ void * r700GetActiveFpShaderBo(GLcontext * ctx) GLboolean r700SetupFragmentProgram(GLcontext * ctx) { context_t *context = R700_CONTEXT(ctx); - BATCH_LOCALS(&context->radeon); R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw); - + BATCH_LOCALS(&context->radeon); struct r700_fragment_program *fp = (struct r700_fragment_program *) (ctx->FragmentProgram._Current); r700_AssemblerBase *pAsm = &(fp->r700AsmCode); @@ -305,6 +311,16 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx) ui = (r700->SPI_PS_IN_CONTROL_0.u32All & NUM_INTERP_mask) / (1 << NUM_INTERP_shift); + /* PS uses fragment.position */ + if (mesa_fp->Base.InputsRead & (1 << FRAG_ATTRIB_WPOS)) + { + ui += 1; + SETfield(r700->SPI_PS_IN_CONTROL_0.u32All, ui, NUM_INTERP_shift, NUM_INTERP_mask); + SETfield(r700->SPI_PS_IN_CONTROL_0.u32All, CENTERS_ONLY, BARYC_SAMPLE_CNTL_shift, BARYC_SAMPLE_CNTL_mask); + SETbit(r700->SPI_PS_IN_CONTROL_0.u32All, POSITION_ENA_bit); + SETbit(r700->SPI_INPUT_Z.u32All, PROVIDE_Z_TO_SPI_bit); + } + ui = (unNumOfReg < ui) ? ui : unNumOfReg; SETfield(r700->ps.SQ_PGM_RESOURCES_PS.u32All, ui, NUM_GPRS_shift, NUM_GPRS_mask); @@ -339,7 +355,6 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx) } /* sent out shader constants. */ - paramList = fp->mesa_program.Base.Parameters; if(NULL != paramList) @@ -349,7 +364,7 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx) unNumParamData = paramList->NumParameters * 4; BEGIN_BATCH_NO_AUTOSTATE(2 + unNumParamData); - + R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_ALU_CONST, unNumParamData)); /* assembler map const from very beginning. */ @@ -369,6 +384,19 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx) } // emit ps input map + unBit = 1 << FRAG_ATTRIB_WPOS; + if(mesa_fp->Base.InputsRead & unBit) + { + ui = pAsm->uiFP_AttributeMap[FRAG_ATTRIB_WPOS]; + SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, SEL_CENTROID_bit); + SETfield(r700->SPI_PS_INPUT_CNTL[ui].u32All, ui, + SEMANTIC_shift, SEMANTIC_mask); + if (r700->SPI_INTERP_CONTROL_0.u32All & FLAT_SHADE_ENA_bit) + SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit); + else + CLEARbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit); + } + unBit = 1 << FRAG_ATTRIB_COL0; if(mesa_fp->Base.InputsRead & unBit) { @@ -424,5 +452,3 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx) return GL_TRUE; } - - diff --git a/src/mesa/drivers/dri/r600/r700_ioctl.c b/src/mesa/drivers/dri/r600/r700_ioctl.c index c4795320012..23cc128d6db 100644 --- a/src/mesa/drivers/dri/r600/r700_ioctl.c +++ b/src/mesa/drivers/dri/r600/r700_ioctl.c @@ -43,7 +43,6 @@ static void r700Flush(GLcontext *ctx) { radeonContextPtr radeon = RADEON_CONTEXT(ctx); - context_t * context = R700_CONTEXT(ctx); if (RADEON_DEBUG & DEBUG_IOCTL) fprintf(stderr, "%s %d\n", __FUNCTION__, radeon->cmdbuf.cs->cdw); @@ -59,7 +58,7 @@ static void r700Flush(GLcontext *ctx) if (radeon->dma.flush) radeon->dma.flush( ctx ); - r700SendContextStates(context); + r700EmitState(ctx); if (radeon->cmdbuf.cs->cdw) rcommonFlushCmdBuf(radeon, __FUNCTION__); diff --git a/src/mesa/drivers/dri/r600/r700_oglprog.c b/src/mesa/drivers/dri/r600/r700_oglprog.c index 36de143b1a7..c49b90c1cc9 100644 --- a/src/mesa/drivers/dri/r600/r700_oglprog.c +++ b/src/mesa/drivers/dri/r600/r700_oglprog.c @@ -33,6 +33,7 @@ #include "tnl/tnl.h" #include "r600_context.h" +#include "r600_emit.h" #include "r700_oglprog.h" #include "r700_fragprog.h" @@ -87,7 +88,6 @@ static void r700DeleteProgram(GLcontext * ctx, struct gl_program *prog) { struct r700_vertex_program * vp; struct r700_fragment_program * fp; - context_t *context = R700_CONTEXT(ctx); switch (prog->Target) { diff --git a/src/mesa/drivers/dri/r600/r700_render.c b/src/mesa/drivers/dri/r600/r700_render.c index 4e0d5391d0a..6705dbcf4b9 100644 --- a/src/mesa/drivers/dri/r600/r700_render.c +++ b/src/mesa/drivers/dri/r600/r700_render.c @@ -44,7 +44,6 @@ #include "tnl/t_vertex.h" #include "tnl/t_pipeline.h" -#include "radeon_mipmap_tree.h" #include "r600_context.h" #include "r600_cmdbuf.h" @@ -58,7 +57,7 @@ void r700WaitForIdle(context_t *context); void r700WaitForIdleClean(context_t *context); void r700Start3D(context_t *context); GLboolean r700SendTextureState(context_t *context); -unsigned int r700PrimitiveType(int prim); +static unsigned int r700PrimitiveType(int prim); void r600UpdateTextureState(GLcontext * ctx); GLboolean r700SyncSurf(context_t *context, struct radeon_bo *pbo, @@ -138,68 +137,15 @@ static GLboolean r700SetupShaders(GLcontext * ctx) exportCount = (r700->ps.SQ_PGM_EXPORTS_PS.u32All & EXPORT_MODE_mask) / (1 << EXPORT_MODE_shift); r700->CB_SHADER_CONTROL.u32All = (1 << exportCount) - 1; - return GL_TRUE; -} + r600UpdateTextureState(ctx); -GLboolean r700SendTextureState(context_t *context) -{ - unsigned int i; - R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw); - offset_modifiers offset_mod = {NO_SHIFT, 0, 0xFFFFFFFF}; - struct radeon_bo *bo = NULL; - BATCH_LOCALS(&context->radeon); + r700SendFSState(context); // FIXME just a place holder for now + r700SendPSState(context); + r700SendVSState(context); + + r700SendTextureState(context); + r700SetupStreams(ctx); - for (i=0; i<R700_TEXTURE_NUMBERUNITS; i++) { - radeonTexObj *t = r700->textures[i]; - if (t) { - if (!t->image_override) - bo = t->mt->bo; - else - bo = t->bo; - if (bo) { - - r700SyncSurf(context, bo, - RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, - 0, TC_ACTION_ENA_bit); - - BEGIN_BATCH_NO_AUTOSTATE(9); - R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7)); - R600_OUT_BATCH(i * 7); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE0); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE1); - R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE2, - bo, - 0, - RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0, &offset_mod); - R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE3, - bo, - 0, - RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0, &offset_mod); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE4); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE5); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE6); - END_BATCH(); - - BEGIN_BATCH_NO_AUTOSTATE(5); - R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_SAMPLER, 3)); - R600_OUT_BATCH(i * 3); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER0); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER1); - R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER2); - END_BATCH(); - - BEGIN_BATCH_NO_AUTOSTATE(2 + 4); - R600_OUT_BATCH_REGSEQ((TD_PS_SAMPLER0_BORDER_RED + (i * 16)), 4); - R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_RED); - R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_GREEN); - R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_BLUE); - R600_OUT_BATCH(r700->textures[i]->TD_PS_SAMPLER0_BORDER_ALPHA); - END_BATCH(); - - COMMIT_BATCH(); - } - } - } return GL_TRUE; } @@ -211,17 +157,15 @@ GLboolean r700SyncSurf(context_t *context, { BATCH_LOCALS(&context->radeon); uint32_t cp_coher_size; - offset_modifiers offset_mod; + + if (!pbo) + return GL_FALSE; if (pbo->size == 0xffffffff) cp_coher_size = 0xffffffff; else cp_coher_size = ((pbo->size + 255) >> 8); - offset_mod.shift = NO_SHIFT; - offset_mod.shiftbits = 0; - offset_mod.mask = 0xFFFFFFFF; - BEGIN_BATCH_NO_AUTOSTATE(5); R600_OUT_BATCH(CP_PACKET3(R600_IT_SURFACE_SYNC, 3)); R600_OUT_BATCH(sync_type); @@ -229,7 +173,7 @@ GLboolean r700SyncSurf(context_t *context, R600_OUT_BATCH_RELOC(0, pbo, 0, - read_domain, write_domain, 0, &offset_mod); // ??? + read_domain, write_domain, 0); // ??? R600_OUT_BATCH(10); END_BATCH(); @@ -238,7 +182,7 @@ GLboolean r700SyncSurf(context_t *context, return GL_TRUE; } -unsigned int r700PrimitiveType(int prim) +static unsigned int r700PrimitiveType(int prim) { switch (prim & PRIM_MODE_MASK) { @@ -279,112 +223,110 @@ unsigned int r700PrimitiveType(int prim) } } -static GLboolean r700RunRender(GLcontext * ctx, - struct tnl_pipeline_stage *stage) +static void r700RunRenderPrimitive(GLcontext * ctx, int start, int end, int prim) { - context_t *context = R700_CONTEXT(ctx); - R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw); - int lastIndex = 0; - BATCH_LOCALS(&context->radeon); + context_t *context = R700_CONTEXT(ctx); + BATCH_LOCALS(&context->radeon); + int type, i, total_emit; + int num_indices = end - start; + uint32_t vgt_draw_initiator = 0; + uint32_t vgt_index_type = 0; + uint32_t vgt_primitive_type = 0; + uint32_t vgt_num_indices = 0; - unsigned int i, j; - TNLcontext *tnl = TNL_CONTEXT(ctx); - struct vertex_buffer *vb = &tnl->vb; - - r700Start3D(context); /* TODO : this is too much. */ - - r700SendSQConfig(context); - - r700UpdateShaders(ctx); - - r700SetScissor(context); - r700SetRenderTarget(context, 0); - r700SetDepthTarget(context); - - if(r700SetupStreams(ctx)) - { - return GL_TRUE; - } - - r600UpdateTextureState(ctx); - r700SendTextureState(context); - - r700SetupShaders(ctx); - - r700SendFSState(context); // FIXME just a place holder for now - r700SendPSState(context); - r700SendVSState(context); + type = r700PrimitiveType(prim); - r700SendUCPState(context); - r700SendContextStates(context); - r700SendViewportState(context, 0); - r700SendRenderTargetState(context, 0); - r700SendDepthTargetState(context); + if (type < 0 || num_indices <= 0) + return; - /* richard test code */ - for (i = 0; i < vb->PrimitiveCount; i++) - { - GLuint prim = _tnl_translate_prim(&vb->Primitive[i]); - GLuint start = vb->Primitive[i].start; - GLuint end = vb->Primitive[i].start + vb->Primitive[i].count; - GLuint numIndices = vb->Primitive[i].count; - GLuint numEntires; - - unsigned int VGT_DRAW_INITIATOR = 0; - unsigned int VGT_INDEX_TYPE = 0; - unsigned int VGT_PRIMITIVE_TYPE = 0; - unsigned int VGT_NUM_INDICES = 0; - - if (numIndices < 1) - continue; - - numEntires = 3 /* VGT_PRIMITIVE_TYPE */ + total_emit = 3 /* VGT_PRIMITIVE_TYPE */ + 2 /* VGT_INDEX_TYPE */ + 2 /* NUM_INSTANCES */ - + numIndices + 3; /* DRAW_INDEX_IMMD */ - - BEGIN_BATCH_NO_AUTOSTATE(numEntires); + + num_indices + 3; /* DRAW_INDEX_IMMD */ + BEGIN_BATCH_NO_AUTOSTATE(total_emit); // prim - VGT_PRIMITIVE_TYPE |= r700PrimitiveType(prim) << VGT_PRIMITIVE_TYPE__PRIM_TYPE_shift; + SETfield(vgt_primitive_type, type, + VGT_PRIMITIVE_TYPE__PRIM_TYPE_shift, VGT_PRIMITIVE_TYPE__PRIM_TYPE_mask); R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); R600_OUT_BATCH(mmVGT_PRIMITIVE_TYPE - ASIC_CONFIG_BASE_INDEX); - R600_OUT_BATCH(VGT_PRIMITIVE_TYPE); + R600_OUT_BATCH(vgt_primitive_type); // index type - VGT_INDEX_TYPE |= DI_INDEX_SIZE_32_BIT << INDEX_TYPE_shift; + SETfield(vgt_index_type, DI_INDEX_SIZE_32_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask); R600_OUT_BATCH(CP_PACKET3(R600_IT_INDEX_TYPE, 0)); - R600_OUT_BATCH(VGT_INDEX_TYPE); + R600_OUT_BATCH(vgt_index_type); // num instances R600_OUT_BATCH(CP_PACKET3(R600_IT_NUM_INSTANCES, 0)); R600_OUT_BATCH(1); // draw packet - VGT_NUM_INDICES = numIndices; - VGT_DRAW_INITIATOR |= DI_SRC_SEL_IMMEDIATE << SOURCE_SELECT_shift; - VGT_DRAW_INITIATOR |= DI_MAJOR_MODE_0 << MAJOR_MODE_shift; + vgt_num_indices = num_indices; + SETfield(vgt_draw_initiator, DI_SRC_SEL_IMMEDIATE, SOURCE_SELECT_shift, SOURCE_SELECT_mask); + SETfield(vgt_draw_initiator, DI_MAJOR_MODE_0, MAJOR_MODE_shift, MAJOR_MODE_mask); - R600_OUT_BATCH(CP_PACKET3(R600_IT_DRAW_INDEX_IMMD, (numIndices + 1))); - R600_OUT_BATCH(VGT_NUM_INDICES); - R600_OUT_BATCH(VGT_DRAW_INITIATOR); + R600_OUT_BATCH(CP_PACKET3(R600_IT_DRAW_INDEX_IMMD, (num_indices + 1))); + R600_OUT_BATCH(vgt_num_indices); + R600_OUT_BATCH(vgt_draw_initiator); - for (j = lastIndex; j < lastIndex + numIndices; j++) - { - R600_OUT_BATCH(j); + for (i = start; i < end; i++) { + R600_OUT_BATCH(i); } - lastIndex += numIndices; - END_BATCH(); COMMIT_BATCH(); + +} + +void r700EmitState(GLcontext * ctx) +{ + context_t *context = R700_CONTEXT(ctx); + radeonContextPtr radeon = &context->radeon; + + if (radeon->cmdbuf.cs->cdw && !radeon->hw.is_dirty && !radeon->hw.all_dirty) + return; + + rcommonEnsureCmdBufSpace(&context->radeon, + context->radeon.hw.max_state_size, __FUNCTION__); + + r700SendSQConfig(context); + + r700SendUCPState(context); + r700SendContextStates(context); + r700SendViewportState(context, 0); + r700SendRenderTargetState(context, 0); + r700SendDepthTargetState(context); + +} + +static GLboolean r700RunRender(GLcontext * ctx, + struct tnl_pipeline_stage *stage) +{ + context_t *context = R700_CONTEXT(ctx); + unsigned int i; + TNLcontext *tnl = TNL_CONTEXT(ctx); + struct vertex_buffer *vb = &tnl->vb; + + r700Start3D(context); + + r700UpdateShaders(ctx); + r700SetScissor(context); + r700SetupShaders(ctx); + + r700EmitState(ctx); + + /* richard test code */ + for (i = 0; i < vb->PrimitiveCount; i++) { + GLuint prim = _tnl_translate_prim(&vb->Primitive[i]); + GLuint start = vb->Primitive[i].start; + GLuint end = vb->Primitive[i].start + vb->Primitive[i].count; + r700RunRenderPrimitive(ctx, start, end, prim); } /* Flush render op cached for last several quads. */ r700WaitForIdleClean(context); - radeonReleaseArrays(ctx, 0); - - rcommonFlushCmdBuf( &context->radeon, __FUNCTION__ ); + radeonReleaseArrays(ctx, ~0); return GL_FALSE; } diff --git a/src/mesa/drivers/dri/r600/r700_state.c b/src/mesa/drivers/dri/r600/r700_state.c index 87ea1719c49..835b5e18c2c 100644 --- a/src/mesa/drivers/dri/r600/r700_state.c +++ b/src/mesa/drivers/dri/r600/r700_state.c @@ -60,6 +60,8 @@ static void r700SetClipPlaneState(GLcontext * ctx, GLenum cap, GLboolean state); static void r700UpdatePolygonMode(GLcontext * ctx); static void r700SetPolygonOffsetState(GLcontext * ctx, GLboolean state); static void r700SetStencilState(GLcontext * ctx, GLboolean state); +static void r700SetRenderTarget(context_t *context, int id); +static void r700SetDepthTarget(context_t *context); void r700SetDefaultStates(context_t *context) //-------------------- { @@ -158,29 +160,16 @@ void r700UpdateViewportOffset(GLcontext * ctx) //------------------ */ void r700UpdateDrawBuffer(GLcontext * ctx) /* TODO */ //--------------------- { -#if 0 /* to be enabled */ - context_t *context = R700_CONTEXT(ctx); + context_t *context = R700_CONTEXT(ctx); - switch (ctx->DrawBuffer->_ColorDrawBufferIndexes[0]) - { - case BUFFER_FRONT_LEFT: - context->target.rt = context->screen->frontBuffer; - break; - case BUFFER_BACK_LEFT: - context->target.rt = context->screen->backBuffer; - break; - default: - memset (&context->target.rt, sizeof(context->target.rt), 0); - } -#endif /* to be enabled */ + r700SetRenderTarget(context, 0); + r700SetDepthTarget(context); } static void r700FetchStateParameter(GLcontext * ctx, const gl_state_index state[STATE_LENGTH], GLfloat * value) { - context_t *context = R700_CONTEXT(ctx); - /* TODO */ } @@ -600,14 +589,46 @@ static void r700BlendFuncSeparate(GLcontext * ctx, /** * Translate LogicOp enums into hardware representation. - * Both use a very logical bit-wise layout, but unfortunately the order - * of bits is reversed. */ static GLuint translate_logicop(GLenum logicop) { - GLuint bits = logicop - GL_CLEAR; - bits = ((bits & 1) << 3) | ((bits & 2) << 1) | ((bits & 4) >> 1) | ((bits & 8) >> 3); - return bits; + switch (logicop) { + case GL_CLEAR: + return 0x00; + case GL_SET: + return 0xff; + case GL_COPY: + return 0xcc; + case GL_COPY_INVERTED: + return 0x33; + case GL_NOOP: + return 0xaa; + case GL_INVERT: + return 0x55; + case GL_AND: + return 0x88; + case GL_NAND: + return 0x77; + case GL_OR: + return 0xee; + case GL_NOR: + return 0x11; + case GL_XOR: + return 0x66; + case GL_EQUIV: + return 0xaa; + case GL_AND_REVERSE: + return 0x44; + case GL_AND_INVERTED: + return 0x22; + case GL_OR_REVERSE: + return 0xdd; + case GL_OR_INVERTED: + return 0xbb; + default: + fprintf(stderr, "unknown blend logic operation %x\n", logicop); + return 0xcc; + } } /** @@ -1327,22 +1348,22 @@ void r700SetScissor(context_t *context) //--------------- r700->viewport[id].enabled = GL_TRUE; } -void r700SetRenderTarget(context_t *context, int id) +static void r700SetRenderTarget(context_t *context, int id) { R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw); struct radeon_renderbuffer *rrb; unsigned int nPitchInPixel; + rrb = radeon_get_colorbuffer(&context->radeon); + if (!rrb || !rrb->bo) { + fprintf(stderr, "no rrb\n"); + return; + } + /* screen/window/view */ SETfield(r700->CB_TARGET_MASK.u32All, 0xF, (4 * id), TARGET0_ENABLE_mask); - rrb = radeon_get_colorbuffer(&context->radeon); - if (!rrb || !rrb->bo) { - fprintf(stderr, "no rrb\n"); - return; - } - /* color buffer */ r700->render_target[id].CB_COLOR0_BASE.u32All = context->radeon.state.color.draw_offset; @@ -1375,39 +1396,22 @@ void r700SetRenderTarget(context_t *context, int id) r700->render_target[id].enabled = GL_TRUE; } -void r700SetDepthTarget(context_t *context) +static void r700SetDepthTarget(context_t *context) { R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw); struct radeon_renderbuffer *rrb; unsigned int nPitchInPixel; + rrb = radeon_get_depthbuffer(&context->radeon); + if (!rrb) + return; + /* depth buf */ r700->DB_DEPTH_SIZE.u32All = 0; r700->DB_DEPTH_BASE.u32All = 0; r700->DB_DEPTH_INFO.u32All = 0; - - r700->DB_DEPTH_CLEAR.u32All = 0x3F800000; - r700->DB_DEPTH_VIEW.u32All = 0; - r700->DB_RENDER_CONTROL.u32All = 0; - SETbit(r700->DB_RENDER_CONTROL.u32All, STENCIL_COMPRESS_DISABLE_bit); - SETbit(r700->DB_RENDER_CONTROL.u32All, DEPTH_COMPRESS_DISABLE_bit); - r700->DB_RENDER_OVERRIDE.u32All = 0; - if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770) - SETbit(r700->DB_RENDER_OVERRIDE.u32All, FORCE_SHADER_Z_ORDER_bit); - SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIZ_ENABLE_shift, FORCE_HIZ_ENABLE_mask); - SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIS_ENABLE0_shift, FORCE_HIS_ENABLE0_mask); - SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIS_ENABLE1_shift, FORCE_HIS_ENABLE1_mask); - - r700->DB_ALPHA_TO_MASK.u32All = 0; - SETfield(r700->DB_ALPHA_TO_MASK.u32All, 2, ALPHA_TO_MASK_OFFSET0_shift, ALPHA_TO_MASK_OFFSET0_mask); - SETfield(r700->DB_ALPHA_TO_MASK.u32All, 2, ALPHA_TO_MASK_OFFSET1_shift, ALPHA_TO_MASK_OFFSET1_mask); - SETfield(r700->DB_ALPHA_TO_MASK.u32All, 2, ALPHA_TO_MASK_OFFSET2_shift, ALPHA_TO_MASK_OFFSET2_mask); - SETfield(r700->DB_ALPHA_TO_MASK.u32All, 2, ALPHA_TO_MASK_OFFSET3_shift, ALPHA_TO_MASK_OFFSET3_mask); - - rrb = radeon_get_depthbuffer(&context->radeon); - if (!rrb) - return; + r700->DB_DEPTH_VIEW.u32All = 0; nPitchInPixel = rrb->pitch/rrb->cpp; @@ -1638,9 +1642,10 @@ static void r700InitSQConfig(GLcontext * ctx) void r700InitState(GLcontext * ctx) //------------------- { context_t *context = R700_CONTEXT(ctx); - R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw); + radeon_firevertices(&context->radeon); + r700->TA_CNTL_AUX.u32All = 0; SETfield(r700->TA_CNTL_AUX.u32All, 28, TD_FIFO_CREDIT_shift, TD_FIFO_CREDIT_mask); r700->VC_ENHANCE.u32All = 0; @@ -1756,6 +1761,24 @@ void r700InitState(GLcontext * ctx) //------------------- r700DepthFunc(ctx, ctx->Depth.Func); SETbit(r700->DB_SHADER_CONTROL.u32All, DUAL_EXPORT_ENABLE_bit); + r700->DB_DEPTH_CLEAR.u32All = 0x3F800000; + + r700->DB_RENDER_CONTROL.u32All = 0; + SETbit(r700->DB_RENDER_CONTROL.u32All, STENCIL_COMPRESS_DISABLE_bit); + SETbit(r700->DB_RENDER_CONTROL.u32All, DEPTH_COMPRESS_DISABLE_bit); + r700->DB_RENDER_OVERRIDE.u32All = 0; + if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770) + SETbit(r700->DB_RENDER_OVERRIDE.u32All, FORCE_SHADER_Z_ORDER_bit); + SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIZ_ENABLE_shift, FORCE_HIZ_ENABLE_mask); + SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIS_ENABLE0_shift, FORCE_HIS_ENABLE0_mask); + SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIS_ENABLE1_shift, FORCE_HIS_ENABLE1_mask); + + r700->DB_ALPHA_TO_MASK.u32All = 0; + SETfield(r700->DB_ALPHA_TO_MASK.u32All, 2, ALPHA_TO_MASK_OFFSET0_shift, ALPHA_TO_MASK_OFFSET0_mask); + SETfield(r700->DB_ALPHA_TO_MASK.u32All, 2, ALPHA_TO_MASK_OFFSET1_shift, ALPHA_TO_MASK_OFFSET1_mask); + SETfield(r700->DB_ALPHA_TO_MASK.u32All, 2, ALPHA_TO_MASK_OFFSET2_shift, ALPHA_TO_MASK_OFFSET2_mask); + SETfield(r700->DB_ALPHA_TO_MASK.u32All, 2, ALPHA_TO_MASK_OFFSET3_shift, ALPHA_TO_MASK_OFFSET3_mask); + /* stencil */ r700Enable(ctx, GL_STENCIL_TEST, ctx->Stencil._Enabled); r700StencilMaskSeparate(ctx, 0, ctx->Stencil.WriteMask[0]); @@ -1821,6 +1844,8 @@ void r700InitState(GLcontext * ctx) //------------------- /* Set up color compare mask */ r700->CB_CLRCMP_MSK.u32All = 0xFFFFFFFF; + context->radeon.hw.all_dirty = GL_TRUE; + } void r700InitStateFuncs(struct dd_function_table *functions) //----------------- diff --git a/src/mesa/drivers/dri/r600/r700_state.h b/src/mesa/drivers/dri/r600/r700_state.h index 23246367db8..30eb54e8b0a 100644 --- a/src/mesa/drivers/dri/r600/r700_state.h +++ b/src/mesa/drivers/dri/r600/r700_state.h @@ -42,10 +42,8 @@ extern void r700UpdateDrawBuffer (GLcontext * ctx); extern void r700InitState (GLcontext * ctx); extern void r700InitStateFuncs (struct dd_function_table *functions); -extern void r700SetRenderTarget(context_t *context, int id); extern void r700SetDefaultStates(context_t * context); -void r700SetScissor(context_t *context); -void r700SetDepthTarget(context_t *context); +extern void r700SetScissor(context_t *context); #endif /* _R600_SCREEN_H */ diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.c b/src/mesa/drivers/dri/r600/r700_vertprog.c index af6a6b8c295..31e71cdfa30 100644 --- a/src/mesa/drivers/dri/r600/r700_vertprog.c +++ b/src/mesa/drivers/dri/r600/r700_vertprog.c @@ -335,17 +335,13 @@ void * r700GetActiveVpShaderBo(GLcontext * ctx) GLboolean r700SetupVertexProgram(GLcontext * ctx) { context_t *context = R700_CONTEXT(ctx); - - BATCH_LOCALS(&context->radeon); - R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw); - + BATCH_LOCALS(&context->radeon); struct r700_vertex_program *vp = (struct r700_vertex_program *)ctx->VertexProgram._Current; struct gl_program_parameter_list *paramList; unsigned int unNumParamData; - unsigned int ui; if(GL_FALSE == vp->loaded) @@ -394,7 +390,6 @@ GLboolean r700SetupVertexProgram(GLcontext * ctx) */ /* sent out shader constants. */ - paramList = vp->mesa_program.Base.Parameters; if(NULL != paramList) @@ -403,8 +398,8 @@ GLboolean r700SetupVertexProgram(GLcontext * ctx) unNumParamData = paramList->NumParameters * 4; - BEGIN_BATCH_NO_AUTOSTATE(unNumParamData + 2); - + BEGIN_BATCH_NO_AUTOSTATE(unNumParamData + 2); + R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_ALU_CONST, unNumParamData)); /* assembler map const from very beginning. */ R600_OUT_BATCH(SQ_ALU_CONSTANT_VS_OFFSET * 4); diff --git a/src/mesa/drivers/dri/radeon/radeon_bo_drm.h b/src/mesa/drivers/dri/radeon/radeon_bo_drm.h index 8eeaea1cb20..d52fb017d8a 100644 --- a/src/mesa/drivers/dri/radeon/radeon_bo_drm.h +++ b/src/mesa/drivers/dri/radeon/radeon_bo_drm.h @@ -83,6 +83,10 @@ struct radeon_bo_funcs { int (*bo_unmap)(struct radeon_bo *bo); int (*bo_wait)(struct radeon_bo *bo); int (*bo_is_static)(struct radeon_bo *bo); + int (*bo_set_tiling)(struct radeon_bo *bo, uint32_t tiling_flags, + uint32_t pitch); + int (*bo_get_tiling)(struct radeon_bo *bo, uint32_t *tiling_flags, + uint32_t *pitch); }; struct radeon_bo_manager { @@ -187,6 +191,18 @@ static inline int _radeon_bo_wait(struct radeon_bo *bo, return bo->bom->funcs->bo_wait(bo); } +static inline int radeon_bo_set_tiling(struct radeon_bo *bo, + uint32_t tiling_flags, uint32_t pitch) +{ + return bo->bom->funcs->bo_set_tiling(bo, tiling_flags, pitch); +} + +static inline int radeon_bo_get_tiling(struct radeon_bo *bo, + uint32_t *tiling_flags, uint32_t *pitch) +{ + return bo->bom->funcs->bo_get_tiling(bo, tiling_flags, pitch); +} + static inline int radeon_bo_is_static(struct radeon_bo *bo) { if (bo->bom->funcs->bo_is_static) diff --git a/src/mesa/drivers/dri/radeon/radeon_bo_legacy.c b/src/mesa/drivers/dri/radeon/radeon_bo_legacy.c index 992eb4611b1..d6d22cb4c37 100644 --- a/src/mesa/drivers/dri/radeon/radeon_bo_legacy.c +++ b/src/mesa/drivers/dri/radeon/radeon_bo_legacy.c @@ -577,6 +577,8 @@ static struct radeon_bo_funcs bo_legacy_funcs = { bo_unmap, NULL, bo_is_static, + NULL, + NULL, }; static int bo_vram_validate(struct radeon_bo *bo, @@ -622,12 +624,34 @@ static int bo_vram_validate(struct radeon_bo *bo, if (bo_legacy->dirty || bo_legacy->tobj->base.dirty_images[0]) { if (IS_R600_CLASS(boml->screen)) { - char *src = bo_legacy->ptr; - char *dst = (char *) boml->screen->driScreen->pFB + - (bo_legacy->offset - boml->fb_location); + drm_radeon_texture_t tex; + drm_radeon_tex_image_t tmp; + int ret; - /* FIXME: alignment, pitch, etc. */ - memcpy(dst, src, bo->size); + tex.offset = bo_legacy->offset; + tex.image = &tmp; + assert(!(tex.offset & 1023)); + + tmp.x = 0; + tmp.y = 0; + tmp.width = bo->size; + tmp.height = 1; + tmp.data = bo_legacy->ptr; + tex.format = RADEON_TXFORMAT_ARGB8888; + tex.width = tmp.width; + tex.height = tmp.height; + tex.pitch = bo->size; + do { + ret = drmCommandWriteRead(bo->bom->fd, + DRM_RADEON_TEXTURE, + &tex, + sizeof(drm_radeon_texture_t)); + if (ret) { + if (RADEON_DEBUG & DEBUG_IOCTL) + fprintf(stderr, "DRM_RADEON_TEXTURE: again!\n"); + usleep(1); + } + } while (ret == -EAGAIN); } else { /* Copy to VRAM using a blit. * All memory is 4K aligned. We're using 1024 pixels wide blits. @@ -904,3 +928,32 @@ unsigned radeon_bo_legacy_relocs_size(struct radeon_bo *bo) return bo->size; } +/* + * Fake up a bo for things like texture image_override. + * bo->offset already includes fb_location + */ +struct radeon_bo *radeon_legacy_bo_alloc_fake(struct radeon_bo_manager *bom, + int size, + uint32_t offset) +{ + struct bo_manager_legacy *boml = (struct bo_manager_legacy *)bom; + struct bo_legacy *bo; + +#ifdef RADEON_DEBUG_BO + bo = bo_allocate(boml, size, 0, RADEON_GEM_DOMAIN_VRAM, 0, "fake bo"); +#else + bo = bo_allocate(boml, size, 0, RADEON_GEM_DOMAIN_VRAM, 0); +#endif /* RADEON_DEBUG_BO */ + if (bo == NULL) + return NULL; + bo->static_bo = 1; + bo->offset = offset; + bo->base.handle = bo->offset; + bo->ptr = boml->screen->driScreen->pFB + (offset - boml->fb_location); + if (bo->base.handle > boml->nhandle) { + boml->nhandle = bo->base.handle + 1; + } + radeon_bo_ref(&(bo->base)); + return &(bo->base); +} + diff --git a/src/mesa/drivers/dri/radeon/radeon_bo_legacy.h b/src/mesa/drivers/dri/radeon/radeon_bo_legacy.h index 0db817cab07..455adebc099 100644 --- a/src/mesa/drivers/dri/radeon/radeon_bo_legacy.h +++ b/src/mesa/drivers/dri/radeon/radeon_bo_legacy.h @@ -42,5 +42,8 @@ struct radeon_bo_manager *radeon_bo_manager_legacy_ctor(struct radeon_screen *sc void radeon_bo_manager_legacy_dtor(struct radeon_bo_manager *bom); void radeon_bo_legacy_texture_age(struct radeon_bo_manager *bom); unsigned radeon_bo_legacy_relocs_size(struct radeon_bo *bo); +struct radeon_bo *radeon_legacy_bo_alloc_fake(struct radeon_bo_manager *bom, + int size, + uint32_t offset); #endif diff --git a/src/mesa/drivers/dri/radeon/radeon_bocs_wrapper.h b/src/mesa/drivers/dri/radeon/radeon_bocs_wrapper.h index e0c70dd9a11..a42870f4a93 100644 --- a/src/mesa/drivers/dri/radeon/radeon_bocs_wrapper.h +++ b/src/mesa/drivers/dri/radeon/radeon_bocs_wrapper.h @@ -15,6 +15,12 @@ #define RADEON_GEM_DOMAIN_GTT 0x2 // GTT or cache flushed #define RADEON_GEM_DOMAIN_VRAM 0x4 // VRAM domain +#define RADEON_TILING_MACRO 0x1 +#define RADEON_TILING_MICRO 0x2 +#define RADEON_TILING_SWAP 0x4 +#define RADEON_TILING_SURFACE 0x8 /* this object requires a surface + * when mapped - i.e. front buffer */ + /* to be used to build locally in mesa with no libdrm bits */ #include "../radeon/radeon_bo_drm.h" #include "../radeon/radeon_cs_drm.h" diff --git a/src/mesa/drivers/dri/radeon/radeon_chipset.h b/src/mesa/drivers/dri/radeon/radeon_chipset.h index 0a6a2df35b8..a275c8fb143 100644 --- a/src/mesa/drivers/dri/radeon/radeon_chipset.h +++ b/src/mesa/drivers/dri/radeon/radeon_chipset.h @@ -356,11 +356,14 @@ #define PCI_CHIP_RV770_947A 0x947A #define PCI_CHIP_RV770_947B 0x947B +#define PCI_CHIP_RV730_9480 0x9480 #define PCI_CHIP_RV730_9487 0x9487 +#define PCI_CHIP_RV730_9488 0x9488 #define PCI_CHIP_RV730_9489 0x9489 #define PCI_CHIP_RV730_948F 0x948F #define PCI_CHIP_RV730_9490 0x9490 #define PCI_CHIP_RV730_9491 0x9491 +#define PCI_CHIP_RV730_9495 0x9495 #define PCI_CHIP_RV730_9498 0x9498 #define PCI_CHIP_RV730_949C 0x949C #define PCI_CHIP_RV730_949E 0x949E @@ -374,12 +377,16 @@ #define PCI_CHIP_RV710_9552 0x9552 #define PCI_CHIP_RV710_9553 0x9553 #define PCI_CHIP_RV710_9555 0x9555 +#define PCI_CHIP_RV710_9557 0x9557 #define PCI_CHIP_RV740_94A0 0x94A0 #define PCI_CHIP_RV740_94A1 0x94A1 +#define PCI_CHIP_RV740_94A3 0x94A3 #define PCI_CHIP_RV740_94B1 0x94B1 #define PCI_CHIP_RV740_94B3 0x94B3 +#define PCI_CHIP_RV740_94B4 0x94B4 #define PCI_CHIP_RV740_94B5 0x94B5 +#define PCI_CHIP_RV740_94B9 0x94B9 enum { CHIP_FAMILY_R100, diff --git a/src/mesa/drivers/dri/radeon/radeon_common.c b/src/mesa/drivers/dri/radeon/radeon_common.c index dde615a4d9d..330c2c8a86e 100644 --- a/src/mesa/drivers/dri/radeon/radeon_common.c +++ b/src/mesa/drivers/dri/radeon/radeon_common.c @@ -481,32 +481,6 @@ void radeonCopyBuffer( __DRIdrawablePrivate *dPriv, if (!n) continue; - if (IS_R600_CLASS(rmesa->radeonScreen)) { - int cpp = rmesa->radeonScreen->cpp; - int src_pitch = rmesa->radeonScreen->backPitch * cpp; - int dst_pitch = rmesa->radeonScreen->frontPitch * cpp; - char *src = (char *)rmesa->radeonScreen->driScreen->pFB + rmesa->radeonScreen->backOffset; - char *dst = (char *)rmesa->radeonScreen->driScreen->pFB + rmesa->radeonScreen->frontOffset; - int j; - drm_clip_rect_t *pb = rmesa->sarea->boxes; - - for (j = 0; j < n; j++) { - int x = pb[j].x1; - int y = pb[j].y1; - int w = pb[j].x2 - x; - int h = pb[j].y2 - y; - - src += (y * src_pitch) + (x * cpp); - dst += (y * dst_pitch) + (x * cpp); - - while (h--) { - memcpy(dst, src, w * cpp); - src += src_pitch; - dst += dst_pitch; - } - } - } - ret = drmCommandNone( rmesa->dri.fd, DRM_RADEON_SWAP ); if ( ret ) { @@ -887,10 +861,11 @@ void radeonUpdatePageFlipping(radeonContextPtr radeon) void radeon_window_moved(radeonContextPtr radeon) { + /* Cliprects has to be updated before doing anything else */ + radeonSetCliprects(radeon); if (!radeon->radeonScreen->driScreen->dri2.enabled) { radeonUpdatePageFlipping(radeon); } - radeonSetCliprects(radeon); } void radeon_viewport(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height) @@ -1092,7 +1067,7 @@ void radeonFlush(GLcontext *ctx) * each of N places that do rendering. This has worse performances, * but it is much easier to get correct. */ - if (radeon->is_front_buffer_rendering) { + if (!radeon->is_front_buffer_rendering) { radeon->front_buffer_dirty = GL_FALSE; } } diff --git a/src/mesa/drivers/dri/radeon/radeon_common_context.c b/src/mesa/drivers/dri/radeon/radeon_common_context.c index a50cd056e1c..f71dc1cb233 100644 --- a/src/mesa/drivers/dri/radeon/radeon_common_context.c +++ b/src/mesa/drivers/dri/radeon/radeon_common_context.c @@ -211,6 +211,7 @@ GLboolean radeonInitContext(radeonContextPtr radeon, radeon->dri.screen = sPriv; radeon->dri.hwContext = driContextPriv->hHWContext; radeon->dri.hwLock = &sPriv->pSAREA->lock; + radeon->dri.hwLockCount = 0; radeon->dri.fd = sPriv->fd; radeon->dri.drmMinor = sPriv->drm_version.minor; @@ -241,7 +242,26 @@ GLboolean radeonInitContext(radeonContextPtr radeon, radeon->texture_depth = ( glVisual->rgbBits > 16 ) ? DRI_CONF_TEXTURE_DEPTH_32 : DRI_CONF_TEXTURE_DEPTH_16; - radeon->texture_row_align = 32; + if (IS_R600_CLASS(radeon->radeonScreen)) { + radeon->texture_row_align = 256; + radeon->texture_rect_row_align = 256; + radeon->texture_compressed_row_align = 256; + } else if (IS_R200_CLASS(radeon->radeonScreen) || + IS_R100_CLASS(radeon->radeonScreen)) { + radeon->texture_row_align = 32; + radeon->texture_rect_row_align = 64; + radeon->texture_compressed_row_align = 32; + } else { /* R300 - not sure this is all correct */ + int chip_family = radeon->radeonScreen->chip_family; + if (chip_family == CHIP_FAMILY_RS600 || + chip_family == CHIP_FAMILY_RS690 || + chip_family == CHIP_FAMILY_RS740) + radeon->texture_row_align = 64; + else + radeon->texture_row_align = 32; + radeon->texture_rect_row_align = 64; + radeon->texture_compressed_row_align = 64; + } return GL_TRUE; } @@ -275,11 +295,10 @@ void radeonDestroyContext(__DRIcontextPrivate *driContextPriv ) GET_CURRENT_CONTEXT(ctx); radeonContextPtr radeon = (radeonContextPtr) driContextPriv->driverPrivate; radeonContextPtr current = ctx ? RADEON_CONTEXT(ctx) : NULL; - - /* +r6/r7 */ - __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv; +#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) /* +r6/r7 */ + __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv; radeonScreenPtr screen = (radeonScreenPtr) (sPriv->private); - /* --------- */ +#endif if (radeon == current) { radeon_firevertices(radeon); @@ -287,16 +306,7 @@ void radeonDestroyContext(__DRIcontextPrivate *driContextPriv ) } assert(radeon); - if (radeon) - { - -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) /* +r6/r7 */ - if (IS_R600_CLASS(screen)) - { - r600DestroyContext(driContextPriv); - } -#endif - + if (radeon) { if (radeon->dma.current) { rcommonFlushCmdBuf( radeon, __FUNCTION__ ); } @@ -743,8 +753,10 @@ radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable) bo = depth_bo; radeon_bo_ref(bo); } else { + uint32_t tiling_flags = 0, pitch = 0; + int ret; #ifdef RADEON_DEBUG_BO - bo = radeon_bo_open(radeon->radeonScreen->bom, + bo = radeon_bo_open(radeon->radeonScreen->bom, buffers[i].name, 0, 0, @@ -765,6 +777,13 @@ radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable) regname, buffers[i].name); } + + ret = radeon_bo_get_tiling(bo, &tiling_flags, &pitch); + if (tiling_flags & RADEON_TILING_MACRO) + bo->flags |= RADEON_BO_FLAGS_MACRO_TILE; + if (tiling_flags & RADEON_TILING_MICRO) + bo->flags |= RADEON_BO_FLAGS_MICRO_TILE; + } if (buffers[i].attachment == __DRI_BUFFER_DEPTH) { diff --git a/src/mesa/drivers/dri/radeon/radeon_common_context.h b/src/mesa/drivers/dri/radeon/radeon_common_context.h index 0cdacb1c361..f8e1a25c9fa 100644 --- a/src/mesa/drivers/dri/radeon/radeon_common_context.h +++ b/src/mesa/drivers/dri/radeon/radeon_common_context.h @@ -365,6 +365,7 @@ struct radeon_dri_mirror { drm_context_t hwContext; drm_hw_lock_t *hwLock; + int hwLockCount; int fd; int drmMinor; }; @@ -429,6 +430,8 @@ struct radeon_context { int texture_depth; float initialMaxAnisotropy; uint32_t texture_row_align; + uint32_t texture_rect_row_align; + uint32_t texture_compressed_row_align; struct radeon_dma dma; struct radeon_hw_state hw; @@ -443,7 +446,6 @@ struct radeon_context { GLuint numClipRects; /* Cliprects for the draw buffer */ drm_clip_rect_t *pClipRects; unsigned int lastStamp; - GLboolean lost_context; drm_radeon_sarea_t *sarea; /* Private SAREA data */ /* Mirrors of some DRI state */ diff --git a/src/mesa/drivers/dri/radeon/radeon_fbo.c b/src/mesa/drivers/dri/radeon/radeon_fbo.c index f28efa33e9a..f05b106aaf8 100644 --- a/src/mesa/drivers/dri/radeon/radeon_fbo.c +++ b/src/mesa/drivers/dri/radeon/radeon_fbo.c @@ -291,7 +291,7 @@ radeon_create_renderbuffer(GLenum format, __DRIdrawablePrivate *driDrawPriv) rrb->base.RedBits = 8; rrb->base.GreenBits = 8; rrb->base.BlueBits = 8; - rrb->base.AlphaBits = 8; + rrb->base.AlphaBits = 0; rrb->base.DataType = GL_UNSIGNED_BYTE; break; case GL_RGBA8: @@ -407,7 +407,7 @@ restart: rrb->cpp = 2; rrb->base._ActualFormat = GL_RGB5; rrb->base._BaseFormat = GL_RGB; - rrb->base.DataType = GL_UNSIGNED_SHORT; + rrb->base.DataType = GL_UNSIGNED_BYTE; DBG("Render to RGB5 texture OK\n"); } else if (texImage->TexFormat == &_mesa_texformat_argb1555) { diff --git a/src/mesa/drivers/dri/radeon/radeon_lock.c b/src/mesa/drivers/dri/radeon/radeon_lock.c index 5774f7ebcf7..6294b7e42be 100644 --- a/src/mesa/drivers/dri/radeon/radeon_lock.c +++ b/src/mesa/drivers/dri/radeon/radeon_lock.c @@ -85,11 +85,35 @@ void radeonGetLock(radeonContextPtr rmesa, GLuint flags) } rmesa->vtbl.get_lock(rmesa); - - rmesa->lost_context = GL_TRUE; } - -void radeon_lock_hardware(radeonContextPtr radeon) +#ifndef NDEBUG +struct lock_debug { + const char* function; + const char* file; + int line; +}; + +static struct lock_debug ldebug = {0}; +#endif + +#if 0 +/** TODO: use atomic operations for reference counting **/ +/** gcc 4.2 has builtin functios for this **/ +#define ATOMIC_INC_AND_FETCH(atomic) __sync_add_and_fetch(&atomic, 1) +#define ATOMIC_DEC_AND_FETCH(atomic) __sync_sub_and_fetch(&atomic, 1) +#else +#define ATOMIC_INC_AND_FETCH(atomic) (++atomic) +#define ATOMIC_DEC_AND_FETCH(atomic) (--atomic) +#endif + + +void radeon_lock_hardware(radeonContextPtr radeon +#ifndef NDEBUG + ,const char* function + ,const char* file + ,const int line +#endif + ) { char ret = 0; struct radeon_framebuffer *rfb = NULL; @@ -104,16 +128,39 @@ void radeon_lock_hardware(radeonContextPtr radeon) } if (!radeon->radeonScreen->driScreen->dri2.enabled) { + if (ATOMIC_INC_AND_FETCH(radeon->dri.hwLockCount) > 1) + { +#ifndef NDEBUG + if ( RADEON_DEBUG & DEBUG_SANITY ) + fprintf(stderr, "*** %d times of recursive call to %s ***\n" + "Original call was from %s (file: %s line: %d)\n" + "Now call is coming from %s (file: %s line: %d)\n" + , radeon->dri.hwLockCount, __FUNCTION__ + , ldebug.function, ldebug.file, ldebug.line + , function, file, line + ); +#endif + return; + } DRM_CAS(radeon->dri.hwLock, radeon->dri.hwContext, (DRM_LOCK_HELD | radeon->dri.hwContext), ret ); if (ret) radeonGetLock(radeon, 0); +#ifndef NDEBUG + ldebug.function = function; + ldebug.file = file; + ldebug.line = line; +#endif } } void radeon_unlock_hardware(radeonContextPtr radeon) { if (!radeon->radeonScreen->driScreen->dri2.enabled) { + if (ATOMIC_DEC_AND_FETCH(radeon->dri.hwLockCount) > 0) + { + return; + } DRM_UNLOCK( radeon->dri.fd, radeon->dri.hwLock, radeon->dri.hwContext ); diff --git a/src/mesa/drivers/dri/radeon/radeon_lock.h b/src/mesa/drivers/dri/radeon/radeon_lock.h index 2817709eed6..da5a5b43715 100644 --- a/src/mesa/drivers/dri/radeon/radeon_lock.h +++ b/src/mesa/drivers/dri/radeon/radeon_lock.h @@ -48,12 +48,22 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. extern void radeonGetLock(radeonContextPtr rmesa, GLuint flags); -void radeon_lock_hardware(radeonContextPtr rmesa); +void radeon_lock_hardware(radeonContextPtr rmesa +#ifndef NDEBUG + ,const char* function + ,const char* file + ,const int line +#endif + ); void radeon_unlock_hardware(radeonContextPtr rmesa); /* Lock the hardware and validate our state. */ +#ifdef NDEBUG #define LOCK_HARDWARE( rmesa ) radeon_lock_hardware(rmesa) +#else +#define LOCK_HARDWARE( rmesa ) radeon_lock_hardware(rmesa, __FUNCTION__, __FILE__, __LINE__) +#endif #define UNLOCK_HARDWARE( rmesa ) radeon_unlock_hardware(rmesa) #endif diff --git a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c index f04a07fecd2..d4082bf68f4 100644 --- a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c +++ b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c @@ -90,16 +90,18 @@ static void compute_tex_image_offset(radeonContextPtr rmesa, radeon_mipmap_tree GLuint face, GLuint level, GLuint* curOffset) { radeon_mipmap_level *lvl = &mt->levels[level]; - uint32_t row_align = rmesa->texture_row_align - 1; + uint32_t row_align; /* Find image size in bytes */ if (mt->compressed) { /* TODO: Is this correct? Need test cases for compressed textures! */ - lvl->rowstride = (lvl->width * mt->bpp + 63) & ~63; + row_align = rmesa->texture_compressed_row_align - 1; + lvl->rowstride = (lvl->width * mt->bpp + row_align) & ~row_align; lvl->size = radeon_compressed_texture_size(mt->radeon->glCtx, lvl->width, lvl->height, lvl->depth, mt->compressed); } else if (mt->target == GL_TEXTURE_RECTANGLE_NV) { - lvl->rowstride = (lvl->width * mt->bpp + 63) & ~63; + row_align = rmesa->texture_rect_row_align - 1; + lvl->rowstride = (lvl->width * mt->bpp + row_align) & ~row_align; lvl->size = lvl->rowstride * lvl->height; } else if (mt->tilebits & RADEON_TXO_MICRO_TILE) { /* tile pattern is 16 bytes x2. mipmaps stay 32 byte aligned, @@ -108,6 +110,7 @@ static void compute_tex_image_offset(radeonContextPtr rmesa, radeon_mipmap_tree lvl->rowstride = (lvl->width * mt->bpp * 2 + 31) & ~31; lvl->size = lvl->rowstride * ((lvl->height + 1) / 2) * lvl->depth; } else { + row_align = rmesa->texture_row_align - 1; lvl->rowstride = (lvl->width * mt->bpp + row_align) & ~row_align; lvl->size = lvl->rowstride * lvl->height * lvl->depth; } @@ -323,7 +326,8 @@ GLboolean radeon_miptree_matches_image(radeon_mipmap_tree *mt, if (face >= mt->faces || level < mt->firstLevel || level > mt->lastLevel) return GL_FALSE; - if (texImage->IsCompressed != mt->compressed) + if ((!texImage->IsCompressed && mt->compressed) || + (texImage->IsCompressed && !mt->compressed)) return GL_FALSE; if (!texImage->IsCompressed && @@ -363,8 +367,8 @@ GLboolean radeon_miptree_matches_texture(radeon_mipmap_tree *mt, struct gl_textu mt->width0 == firstImage->Width && mt->height0 == firstImage->Height && mt->depth0 == firstImage->Depth && - mt->bpp == firstImage->TexFormat->TexelBytes && - mt->compressed == compressed); + mt->compressed == compressed && + (!mt->compressed ? (mt->bpp == firstImage->TexFormat->TexelBytes) : 1)); } diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c index 5f1af5b0da5..7b759661ca7 100644 --- a/src/mesa/drivers/dri/radeon/radeon_screen.c +++ b/src/mesa/drivers/dri/radeon/radeon_screen.c @@ -878,11 +878,14 @@ static int radeon_set_screen_flags(radeonScreenPtr screen, int device_id) screen->chip_flags = RADEON_CHIPSET_TCL; break; + case PCI_CHIP_RV730_9480: case PCI_CHIP_RV730_9487: + case PCI_CHIP_RV730_9488: case PCI_CHIP_RV730_9489: case PCI_CHIP_RV730_948F: case PCI_CHIP_RV730_9490: case PCI_CHIP_RV730_9491: + case PCI_CHIP_RV730_9495: case PCI_CHIP_RV730_9498: case PCI_CHIP_RV730_949C: case PCI_CHIP_RV730_949E: @@ -899,15 +902,19 @@ static int radeon_set_screen_flags(radeonScreenPtr screen, int device_id) case PCI_CHIP_RV710_9552: case PCI_CHIP_RV710_9553: case PCI_CHIP_RV710_9555: + case PCI_CHIP_RV710_9557: screen->chip_family = CHIP_FAMILY_RV710; screen->chip_flags = RADEON_CHIPSET_TCL; break; case PCI_CHIP_RV740_94A0: case PCI_CHIP_RV740_94A1: + case PCI_CHIP_RV740_94A3: case PCI_CHIP_RV740_94B1: case PCI_CHIP_RV740_94B3: + case PCI_CHIP_RV740_94B4: case PCI_CHIP_RV740_94B5: + case PCI_CHIP_RV740_94B9: screen->chip_family = CHIP_FAMILY_RV740; screen->chip_flags = RADEON_CHIPSET_TCL; break; @@ -1579,21 +1586,11 @@ static GLboolean radeonCreateContext(const __GLcontextModes * glVisual, { __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv; radeonScreenPtr screen = (radeonScreenPtr) (sPriv->private); -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) - if (IS_R600_CLASS(screen)) - return r600CreateContext(glVisual, driContextPriv, sharedContextPriv); -#endif - #if RADEON_COMMON && defined(RADEON_COMMON_FOR_R300) if (IS_R300_CLASS(screen)) return r300CreateContext(glVisual, driContextPriv, sharedContextPriv); #endif -#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) - if (IS_R200_CLASS(screen)) - return r200CreateContext(glVisual, driContextPriv, sharedContextPriv); -#endif - #if !RADEON_COMMON (void)screen; return r100CreateContext(glVisual, driContextPriv, sharedContextPriv); @@ -1793,8 +1790,16 @@ getSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo ) const struct __DriverAPIRec driDriverAPI = { .InitScreen = radeonInitScreen, .DestroyScreen = radeonDestroyScreen, +#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200) + .CreateContext = r200CreateContext, + .DestroyContext = r200DestroyContext, +#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R600) + .CreateContext = r600CreateContext, + .DestroyContext = r600DestroyContext, +#else .CreateContext = radeonCreateContext, .DestroyContext = radeonDestroyContext, +#endif .CreateBuffer = radeonCreateBuffer, .DestroyBuffer = radeonDestroyBuffer, .SwapBuffers = radeonSwapBuffers, diff --git a/src/mesa/drivers/dri/radeon/radeon_span.c b/src/mesa/drivers/dri/radeon/radeon_span.c index b2a468b4fd6..5e4bf00d7ab 100644 --- a/src/mesa/drivers/dri/radeon/radeon_span.c +++ b/src/mesa/drivers/dri/radeon/radeon_span.c @@ -51,6 +51,59 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. static void radeonSetSpanFunctions(struct radeon_renderbuffer *rrb); + +/* r200 depth buffer is always tiled - this is the formula + according to the docs unless I typo'ed in it +*/ +static GLubyte *r200_depth_2byte(const struct radeon_renderbuffer * rrb, + GLint x, GLint y) +{ + GLubyte *ptr = rrb->bo->ptr; + GLint offset; + if (rrb->has_surface) { + offset = x * rrb->cpp + y * rrb->pitch; + } else { + GLuint b; + offset = 0; + b = (((y >> 4) * (rrb->pitch >> 8) + (x >> 6))); + offset += (b >> 1) << 12; + offset += (((rrb->pitch >> 8) & 0x1) ? (b & 0x1) : ((b & 0x1) ^ ((y >> 4) & 0x1))) << 11; + offset += ((y >> 2) & 0x3) << 9; + offset += ((x >> 3) & 0x1) << 8; + offset += ((x >> 4) & 0x3) << 6; + offset += ((x >> 2) & 0x1) << 5; + offset += ((y >> 1) & 0x1) << 4; + offset += ((x >> 1) & 0x1) << 3; + offset += (y & 0x1) << 2; + offset += (x & 0x1) << 1; + } + return &ptr[offset]; +} + +static GLubyte *r200_depth_4byte(const struct radeon_renderbuffer * rrb, + GLint x, GLint y) +{ + GLubyte *ptr = rrb->bo->ptr; + GLint offset; + if (rrb->has_surface) { + offset = x * rrb->cpp + y * rrb->pitch; + } else { + GLuint b; + offset = 0; + b = (((y & 0x7ff) >> 4) * (rrb->pitch >> 7) + (x >> 5)); + offset += (b >> 1) << 12; + offset += (((rrb->pitch >> 7) & 0x1) ? (b & 0x1) : ((b & 0x1) ^ ((y >> 4) & 0x1))) << 11; + offset += ((y >> 2) & 0x3) << 9; + offset += ((x >> 2) & 0x1) << 8; + offset += ((x >> 3) & 0x3) << 6; + offset += ((y >> 1) & 0x1) << 5; + offset += ((x >> 1) & 0x1) << 4; + offset += (y & 0x1) << 3; + offset += (x & 0x1) << 2; + } + return &ptr[offset]; +} + /* radeon tiling on r300-r500 has 4 states, macro-linear/micro-linear macro-linear/micro-tiled @@ -61,7 +114,6 @@ static void radeonSetSpanFunctions(struct radeon_renderbuffer *rrb); 4 byte surface 8/16 byte (unused) */ - static GLubyte *radeon_ptr_4byte(const struct radeon_renderbuffer * rrb, GLint x, GLint y) { @@ -285,11 +337,21 @@ s8z24_to_z24s8(uint32_t val) */ #define VALUE_TYPE GLushort +#if defined(RADEON_COMMON_FOR_R200) +#define WRITE_DEPTH( _x, _y, d ) \ + *(GLushort *)r200_depth_2byte(rrb, _x + x_off, _y + y_off) = d +#else #define WRITE_DEPTH( _x, _y, d ) \ *(GLushort *)radeon_ptr_2byte_8x2(rrb, _x + x_off, _y + y_off) = d +#endif +#if defined(RADEON_COMMON_FOR_R200) +#define READ_DEPTH( d, _x, _y ) \ + d = *(GLushort *)r200_depth_2byte(rrb, _x + x_off, _y + y_off) +#else #define READ_DEPTH( d, _x, _y ) \ d = *(GLushort *)radeon_ptr_2byte_8x2(rrb, _x + x_off, _y + y_off) +#endif #define TAG(x) radeon##x##_z16 #include "depthtmp.h" @@ -301,7 +363,7 @@ s8z24_to_z24s8(uint32_t val) */ #define VALUE_TYPE GLuint -#ifdef COMPILE_R300 +#if defined(COMPILE_R300) #define WRITE_DEPTH( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off ); \ @@ -310,6 +372,15 @@ do { \ tmp |= ((d << 8) & 0xffffff00); \ *_ptr = tmp; \ } while (0) +#elif defined(RADEON_COMMON_FOR_R200) +#define WRITE_DEPTH( _x, _y, d ) \ +do { \ + GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off ); \ + GLuint tmp = *_ptr; \ + tmp &= 0xff000000; \ + tmp |= ((d) & 0x00ffffff); \ + *_ptr = tmp; \ +} while (0) #else #define WRITE_DEPTH( _x, _y, d ) \ do { \ @@ -321,19 +392,21 @@ do { \ } while (0) #endif -#ifdef COMPILE_R300 +#if defined(COMPILE_R300) #define READ_DEPTH( d, _x, _y ) \ do { \ d = (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)) & 0xffffff00) >> 8; \ }while(0) +#elif defined(RADEON_COMMON_FOR_R200) +#define READ_DEPTH( d, _x, _y ) \ + do { \ + d = *(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off)) & 0x00ffffff; \ + }while(0) #else #define READ_DEPTH( d, _x, _y ) \ d = *(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)) & 0x00ffffff; #endif -/* - fprintf(stderr, "dval(%d, %d, %d, %d)=0x%08X\n", _x, xo, _y, yo, d);\ - d = *(GLuint*)(radeon_ptr(rrb, _x, _y )) & 0x00ffffff; -*/ + #define TAG(x) radeon##x##_z24 #include "depthtmp.h" @@ -345,12 +418,19 @@ do { \ */ #define VALUE_TYPE GLuint -#ifdef COMPILE_R300 +#if defined(COMPILE_R300) #define WRITE_DEPTH( _x, _y, d ) \ do { \ GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off ); \ *_ptr = d; \ } while (0) +#elif defined(RADEON_COMMON_FOR_R200) +#define WRITE_DEPTH( _x, _y, d ) \ +do { \ + GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off ); \ + GLuint tmp = z24s8_to_s8z24(d); \ + *_ptr = tmp; \ +} while (0) #else #define WRITE_DEPTH( _x, _y, d ) \ do { \ @@ -360,20 +440,22 @@ do { \ } while (0) #endif -#ifdef COMPILE_R300 +#if defined(COMPILE_R300) #define READ_DEPTH( d, _x, _y ) \ do { \ d = (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off))); \ }while(0) +#elif defined(RADEON_COMMON_FOR_R200) +#define READ_DEPTH( d, _x, _y ) \ + do { \ + d = s8z24_to_z24s8(*(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off))); \ + }while(0) #else #define READ_DEPTH( d, _x, _y ) do { \ d = s8z24_to_z24s8(*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off ))); \ } while (0) #endif -/* - fprintf(stderr, "dval(%d, %d, %d, %d)=0x%08X\n", _x, xo, _y, yo, d);\ - d = *(GLuint*)(radeon_ptr(rrb, _x, _y )) & 0x00ffffff; -*/ + #define TAG(x) radeon##x##_z24_s8 #include "depthtmp.h" @@ -392,6 +474,15 @@ do { \ tmp |= (d) & 0xff; \ *_ptr = tmp; \ } while (0) +#elif defined(RADEON_COMMON_FOR_R200) +#define WRITE_STENCIL( _x, _y, d ) \ +do { \ + GLuint *_ptr = (GLuint*)r200_depth_4byte(rrb, _x + x_off, _y + y_off); \ + GLuint tmp = *_ptr; \ + tmp &= 0x00ffffff; \ + tmp |= (((d) & 0xff) << 24); \ + *_ptr = tmp; \ +} while (0) #else #define WRITE_STENCIL( _x, _y, d ) \ do { \ @@ -410,6 +501,13 @@ do { \ GLuint tmp = *_ptr; \ d = tmp & 0x000000ff; \ } while (0) +#elif defined(RADEON_COMMON_FOR_R200) +#define READ_STENCIL( d, _x, _y ) \ +do { \ + GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off ); \ + GLuint tmp = *_ptr; \ + d = (tmp & 0xff000000) >> 24; \ +} while (0) #else #define READ_STENCIL( d, _x, _y ) \ do { \ diff --git a/src/mesa/drivers/dri/radeon/radeon_texture.c b/src/mesa/drivers/dri/radeon/radeon_texture.c index ad501c454ce..fa16f44c18e 100644 --- a/src/mesa/drivers/dri/radeon/radeon_texture.c +++ b/src/mesa/drivers/dri/radeon/radeon_texture.c @@ -610,9 +610,17 @@ static void radeon_teximage( if (pixels) { radeon_teximage_map(image, GL_TRUE); - if (compressed) { - memcpy(texImage->Data, pixels, imageSize); + if (image->mt) { + uint32_t srcRowStride, bytesPerRow, rows; + srcRowStride = _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, width); + bytesPerRow = srcRowStride; + rows = (height + 3) / 4; + copy_rows(texImage->Data, image->mt->levels[level].rowstride, + pixels, srcRowStride, rows, bytesPerRow); + } else { + memcpy(texImage->Data, pixels, imageSize); + } } else { GLuint dstRowStride; GLuint *dstImageOffsets; @@ -756,14 +764,23 @@ static void radeon_texsubimage(GLcontext* ctx, int dims, GLenum target, int leve } if (compressed) { - uint32_t srcRowStride, bytesPerRow, rows; - dstRowStride = _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, texImage->Width); + uint32_t srcRowStride, bytesPerRow, rows; + GLubyte *img_start; + if (!image->mt) { + dstRowStride = _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, texImage->Width); + img_start = _mesa_compressed_image_address(xoffset, yoffset, 0, + texImage->TexFormat->MesaFormat, + texImage->Width, texImage->Data); + } + else { + uint32_t blocks_x = dstRowStride / (image->mt->bpp * 4); + img_start = texImage->Data + image->mt->bpp * 4 * (blocks_x * (yoffset / 4) + xoffset / 4); + } srcRowStride = _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, width); bytesPerRow = srcRowStride; - rows = height / 4; + rows = (height + 3) / 4; - copy_rows(texImage->Data, dstRowStride, image->base.Data, srcRowStride, rows, - bytesPerRow); + copy_rows(img_start, dstRowStride, pixels, srcRowStride, rows, bytesPerRow); } else { if (!texImage->TexFormat->StoreImage(ctx, dims, texImage->_BaseFormat, @@ -884,8 +901,8 @@ static void migrate_image_to_miptree(radeon_mipmap_tree *mt, radeon_texture_imag uint32_t height; /* need to confirm this value is correct */ if (mt->compressed) { - height = image->base.Height / 4; - srcrowstride = image->base.RowStride * mt->bpp; + height = (image->base.Height + 3) / 4; + srcrowstride = _mesa_compressed_row_stride(image->base.TexFormat->MesaFormat, image->base.Width); } else { height = image->base.Height * image->base.Depth; srcrowstride = image->base.Width * image->base.TexFormat->TexelBytes; @@ -951,7 +968,7 @@ int radeon_validate_texture_miptree(GLcontext * ctx, struct gl_texture_object *t fprintf(stderr, " Allocate new miptree\n"); radeon_try_alloc_miptree(rmesa, t, &baseimage->base, 0, texObj->BaseLevel); if (!t->mt) { - _mesa_problem(ctx, "r300_validate_texture failed to alloc miptree"); + _mesa_problem(ctx, "radeon_validate_texture failed to alloc miptree"); return GL_FALSE; } } @@ -1000,6 +1017,8 @@ radeon_get_tex_image(GLcontext * ctx, GLenum target, GLint level, } if (compressed) { + /* FIXME: this can't work for small textures (mips) which + use different hw stride */ _mesa_get_compressed_teximage(ctx, target, level, pixels, texObj, texImage); } else { diff --git a/src/mesa/glapi/EXT_framebuffer_object.xml b/src/mesa/glapi/EXT_framebuffer_object.xml index 1b0de2ad231..5559b48b11a 100644 --- a/src/mesa/glapi/EXT_framebuffer_object.xml +++ b/src/mesa/glapi/EXT_framebuffer_object.xml @@ -192,42 +192,4 @@ </function> </category> - - -<category name="GL_EXT_texture_array" number="329"> - <enum name="TEXTURE_1D_ARRAY_EXT" value="0x8C18"/> - <enum name="PROXY_TEXTURE_1D_ARRAY_EXT" value="0x8C19"/> - <enum name="TEXTURE_2D_ARRAY_EXT" value="0x8C1A"/> - <enum name="PROXY_TEXTURE_2D_ARRAY_EXT" value="0x8C1B"/> - - <enum name="TEXTURE_BINDING_1D_ARRAY_EXT" count="1" value="0x8C1C"> - <size name="Get" mode="get"/> - </enum> - - <enum name="TEXTURE_BINDING_2D_ARRAY_EXT" count="1" value="0x8C1D"> - <size name="Get" mode="get"/> - </enum> - - <enum name="MAX_ARRAY_TEXTURE_LAYERS_EXT" count="1" value="0x88FF"> - <size name="Get" mode="get"/> - </enum> - - <enum name="COMPARE_REF_DEPTH_TO_TEXTURE_EXT" count="1" value="0x884E"> - <size name="Get" mode="get"/> - </enum> - - <enum name="FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT" count="1" value="0x8CD4"> - <size name="GetFramebufferAttachmentParameterivEXT" mode="get"/> - </enum> - - <function name="FramebufferTextureLayerEXT" offset="assign"> - <param name="target" type="GLenum"/> - <param name="attachment" type="GLenum"/> - <param name="texture" type="GLuint"/> - <param name="level" type="GLint"/> - <param name="layer" type="GLint"/> - </function> -</category> - - </OpenGLAPI> diff --git a/src/mesa/glapi/EXT_texture_array.xml b/src/mesa/glapi/EXT_texture_array.xml new file mode 100644 index 00000000000..e5bd9f3c697 --- /dev/null +++ b/src/mesa/glapi/EXT_texture_array.xml @@ -0,0 +1,41 @@ +<?xml version="1.0"?> +<!DOCTYPE OpenGLAPI SYSTEM "gl_API.dtd"> + +<OpenGLAPI> + +<category name="GL_EXT_texture_array" number="329"> + <enum name="TEXTURE_1D_ARRAY_EXT" value="0x8C18"/> + <enum name="PROXY_TEXTURE_1D_ARRAY_EXT" value="0x8C19"/> + <enum name="TEXTURE_2D_ARRAY_EXT" value="0x8C1A"/> + <enum name="PROXY_TEXTURE_2D_ARRAY_EXT" value="0x8C1B"/> + + <enum name="TEXTURE_BINDING_1D_ARRAY_EXT" count="1" value="0x8C1C"> + <size name="Get" mode="get"/> + </enum> + + <enum name="TEXTURE_BINDING_2D_ARRAY_EXT" count="1" value="0x8C1D"> + <size name="Get" mode="get"/> + </enum> + + <enum name="MAX_ARRAY_TEXTURE_LAYERS_EXT" count="1" value="0x88FF"> + <size name="Get" mode="get"/> + </enum> + + <enum name="COMPARE_REF_DEPTH_TO_TEXTURE_EXT" count="1" value="0x884E"> + <size name="Get" mode="get"/> + </enum> + + <enum name="FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT" count="1" value="0x8CD4"> + <size name="GetFramebufferAttachmentParameterivEXT" mode="get"/> + </enum> + + <function name="FramebufferTextureLayerEXT" offset="assign"> + <param name="target" type="GLenum"/> + <param name="attachment" type="GLenum"/> + <param name="texture" type="GLuint"/> + <param name="level" type="GLint"/> + <param name="layer" type="GLint"/> + </function> +</category> + +</OpenGLAPI> diff --git a/src/mesa/glapi/gl_API.xml b/src/mesa/glapi/gl_API.xml index 06560d175af..1703637c7b4 100644 --- a/src/mesa/glapi/gl_API.xml +++ b/src/mesa/glapi/gl_API.xml @@ -12255,6 +12255,65 @@ </function> </category> +<category name="GL_APPLE_flush_buffer_range" number="321"> + <enum name="BUFFER_SERIALIZED_MODIFY_APPLE" count="1" value="0x8A12"> + <size name="GetBufferParameteriv" mode="get"/> + </enum> + <enum name="BUFFER_FLUSHING_UNMAP_APPLE" count="1" value="0x8A13"> + <size name="GetBufferParameteriv" mode="get"/> + </enum> + <function name="BufferParameteriAPPLE" offset="assign" static_dispatch="false"> + <param name="target" type="GLenum"/> + <param name="pname" type="GLenum"/> + <param name="param" type="GLint"/> + </function> + <function name="FlushMappedBufferRangeAPPLE" offset="assign" static_dispatch="false"> + <param name="target" type="GLenum"/> + <param name="offset" type="GLintptr"/> + <param name="size" type="GLsizeiptr"/> + </function> +</category> + +<xi:include href="EXT_texture_array.xml" xmlns:xi="http://www.w3.org/2001/XInclude"/> + +<category name="GL_APPLE_texture_range" number="367"> + <enum name="TEXTURE_STORAGE_HINT_APPLE" count="1" value="0x85BC"> + <size name="TexParameteriv"/> + <size name="TexParameterfv"/> + <size name="GetTexParameteriv" mode="get"/> + <size name="GetTexParameterfv" mode="get"/> + </enum> + <enum name="STORAGE_PRIVATE_APPLE" count="1" value="0x85BD"> + <size name="TexParameteriv"/> + <size name="TexParameterfv"/> + </enum> + <enum name="STORAGE_CACHED_APPLE" count="1" value="0x85BE"> + <size name="TexParameteriv"/> + <size name="TexParameterfv"/> + </enum> + <enum name="STORAGE_SHARED_APPLE" count="1" value="0x85BF"> + <size name="TexParameteriv"/> + <size name="TexParameterfv"/> + </enum> + <enum name="TEXTURE_RANGE_LENGTH_APPLE" count="1" value="0x85B7"> + <size name="GetTexParameteriv" mode="get"/> + <size name="GetTexParameterfv" mode="get"/> + </enum> + <enum name="TEXTURE_RANGE_POINTER_APPLE" count="1" value="0x85B8"> + <size name="GetTexParameterPointervAPPLE" mode="get"/> + </enum> + <function name="TextureRangeAPPLE" offset="assign" static_dispatch="false"> + <param name="target" type="GLenum"/> + <param name="length" type="GLsizei"/> + <param name="pointer" type="GLvoid *"/> + </function> + <function name="GetTexParameterPointervAPPLE" offset="assign" static_dispatch="false"> + <param name="target" type="GLenum"/> + <param name="pname" type="GLenum"/> + <param name="params" type="GLvoid **"/> + </function> +</category> + <!-- Unnumbered extensions sorted by name. --> <category name="GL_ATI_blend_equation_separate"> diff --git a/src/mesa/main/api_arrayelt.c b/src/mesa/main/api_arrayelt.c index f5b7d1e1385..2462a1b0037 100644 --- a/src/mesa/main/api_arrayelt.c +++ b/src/mesa/main/api_arrayelt.c @@ -28,6 +28,7 @@ #include "glheader.h" #include "api_arrayelt.h" +#include "bufferobj.h" #include "context.h" #include "imports.h" #include "macros.h" @@ -1071,7 +1072,7 @@ void _ae_destroy_context( GLcontext *ctx ) static void check_vbo( AEcontext *actx, struct gl_buffer_object *vbo ) { - if (vbo->Name && !vbo->Pointer) { + if (_mesa_is_bufferobj(vbo) && !_mesa_bufferobj_mapped(vbo)) { GLuint i; for (i = 0; i < actx->nr_vbos; i++) if (actx->vbo[i] == vbo) diff --git a/src/mesa/main/api_validate.c b/src/mesa/main/api_validate.c index b2f11ffbfe3..33f4dd152a7 100644 --- a/src/mesa/main/api_validate.c +++ b/src/mesa/main/api_validate.c @@ -24,6 +24,7 @@ #include "glheader.h" #include "api_validate.h" +#include "bufferobj.h" #include "context.h" #include "imports.h" #include "mtypes.h" @@ -62,7 +63,7 @@ max_buffer_index(GLcontext *ctx, GLuint count, GLenum type, GLuint max = 0; GLuint i; - if (elementBuf->Name) { + if (_mesa_is_bufferobj(elementBuf)) { /* elements are in a user-defined buffer object. need to map it */ map = ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, GL_READ_ONLY, elementBuf); @@ -164,7 +165,7 @@ _mesa_validate_DrawElements(GLcontext *ctx, return GL_FALSE; /* Vertex buffer object tests */ - if (ctx->Array.ElementArrayBufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Array.ElementArrayBufferObj)) { /* use indices in the buffer object */ /* make sure count doesn't go outside buffer bounds */ if (index_bytes(type, count) > ctx->Array.ElementArrayBufferObj->Size) { @@ -237,7 +238,7 @@ _mesa_validate_DrawRangeElements(GLcontext *ctx, GLenum mode, return GL_FALSE; /* Vertex buffer object tests */ - if (ctx->Array.ElementArrayBufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Array.ElementArrayBufferObj)) { /* use indices in the buffer object */ /* make sure count doesn't go outside buffer bounds */ if (index_bytes(type, count) > ctx->Array.ElementArrayBufferObj->Size) { diff --git a/src/mesa/main/attrib.c b/src/mesa/main/attrib.c index cb49c4cb076..ab99ca1c642 100644 --- a/src/mesa/main/attrib.c +++ b/src/mesa/main/attrib.c @@ -174,24 +174,30 @@ struct texture_state /** - * Allocate a new attribute state node. These nodes have a - * "kind" value and a pointer to a struct of state data. + * Allocate new attribute node of given type/kind. Attach payload data. + * Insert it into the linked list named by 'head'. */ -static struct gl_attrib_node * -new_attrib_node( GLbitfield kind ) +static void +save_attrib_data(struct gl_attrib_node **head, + GLbitfield kind, void *payload) { - struct gl_attrib_node *an = MALLOC_STRUCT(gl_attrib_node); - if (an) { - an->kind = kind; + struct gl_attrib_node *n = MALLOC_STRUCT(gl_attrib_node); + if (n) { + n->kind = kind; + n->data = payload; + /* insert at head */ + n->next = *head; + *head = n; + } + else { + /* out of memory! */ } - return an; } void GLAPIENTRY _mesa_PushAttrib(GLbitfield mask) { - struct gl_attrib_node *newnode; struct gl_attrib_node *head; GET_CURRENT_CONTEXT(ctx); @@ -213,10 +219,7 @@ _mesa_PushAttrib(GLbitfield mask) struct gl_accum_attrib *attr; attr = MALLOC_STRUCT( gl_accum_attrib ); MEMCPY( attr, &ctx->Accum, sizeof(struct gl_accum_attrib) ); - newnode = new_attrib_node( GL_ACCUM_BUFFER_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_ACCUM_BUFFER_BIT, attr); } if (mask & GL_COLOR_BUFFER_BIT) { @@ -227,10 +230,7 @@ _mesa_PushAttrib(GLbitfield mask) /* push the Draw FBO's DrawBuffer[] state, not ctx->Color.DrawBuffer[] */ for (i = 0; i < ctx->Const.MaxDrawBuffers; i ++) attr->DrawBuffer[i] = ctx->DrawBuffer->ColorDrawBuffer[i]; - newnode = new_attrib_node( GL_COLOR_BUFFER_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_COLOR_BUFFER_BIT, attr); } if (mask & GL_CURRENT_BIT) { @@ -238,20 +238,14 @@ _mesa_PushAttrib(GLbitfield mask) FLUSH_CURRENT( ctx, 0 ); attr = MALLOC_STRUCT( gl_current_attrib ); MEMCPY( attr, &ctx->Current, sizeof(struct gl_current_attrib) ); - newnode = new_attrib_node( GL_CURRENT_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_CURRENT_BIT, attr); } if (mask & GL_DEPTH_BUFFER_BIT) { struct gl_depthbuffer_attrib *attr; attr = MALLOC_STRUCT( gl_depthbuffer_attrib ); MEMCPY( attr, &ctx->Depth, sizeof(struct gl_depthbuffer_attrib) ); - newnode = new_attrib_node( GL_DEPTH_BUFFER_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_DEPTH_BUFFER_BIT, attr); } if (mask & GL_ENABLE_BIT) { @@ -331,40 +325,28 @@ _mesa_PushAttrib(GLbitfield mask) attr->VertexProgram = ctx->VertexProgram.Enabled; attr->VertexProgramPointSize = ctx->VertexProgram.PointSizeEnabled; attr->VertexProgramTwoSide = ctx->VertexProgram.TwoSideEnabled; - newnode = new_attrib_node( GL_ENABLE_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_ENABLE_BIT, attr); } if (mask & GL_EVAL_BIT) { struct gl_eval_attrib *attr; attr = MALLOC_STRUCT( gl_eval_attrib ); MEMCPY( attr, &ctx->Eval, sizeof(struct gl_eval_attrib) ); - newnode = new_attrib_node( GL_EVAL_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_EVAL_BIT, attr); } if (mask & GL_FOG_BIT) { struct gl_fog_attrib *attr; attr = MALLOC_STRUCT( gl_fog_attrib ); MEMCPY( attr, &ctx->Fog, sizeof(struct gl_fog_attrib) ); - newnode = new_attrib_node( GL_FOG_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_FOG_BIT, attr); } if (mask & GL_HINT_BIT) { struct gl_hint_attrib *attr; attr = MALLOC_STRUCT( gl_hint_attrib ); MEMCPY( attr, &ctx->Hint, sizeof(struct gl_hint_attrib) ); - newnode = new_attrib_node( GL_HINT_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_HINT_BIT, attr); } if (mask & GL_LIGHTING_BIT) { @@ -372,30 +354,21 @@ _mesa_PushAttrib(GLbitfield mask) FLUSH_CURRENT(ctx, 0); /* flush material changes */ attr = MALLOC_STRUCT( gl_light_attrib ); MEMCPY( attr, &ctx->Light, sizeof(struct gl_light_attrib) ); - newnode = new_attrib_node( GL_LIGHTING_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_LIGHTING_BIT, attr); } if (mask & GL_LINE_BIT) { struct gl_line_attrib *attr; attr = MALLOC_STRUCT( gl_line_attrib ); MEMCPY( attr, &ctx->Line, sizeof(struct gl_line_attrib) ); - newnode = new_attrib_node( GL_LINE_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_LINE_BIT, attr); } if (mask & GL_LIST_BIT) { struct gl_list_attrib *attr; attr = MALLOC_STRUCT( gl_list_attrib ); MEMCPY( attr, &ctx->List, sizeof(struct gl_list_attrib) ); - newnode = new_attrib_node( GL_LIST_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_LIST_BIT, attr); } if (mask & GL_PIXEL_MODE_BIT) { @@ -404,60 +377,42 @@ _mesa_PushAttrib(GLbitfield mask) MEMCPY( attr, &ctx->Pixel, sizeof(struct gl_pixel_attrib) ); /* push the Read FBO's ReadBuffer state, not ctx->Pixel.ReadBuffer */ attr->ReadBuffer = ctx->ReadBuffer->ColorReadBuffer; - newnode = new_attrib_node( GL_PIXEL_MODE_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_PIXEL_MODE_BIT, attr); } if (mask & GL_POINT_BIT) { struct gl_point_attrib *attr; attr = MALLOC_STRUCT( gl_point_attrib ); MEMCPY( attr, &ctx->Point, sizeof(struct gl_point_attrib) ); - newnode = new_attrib_node( GL_POINT_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_POINT_BIT, attr); } if (mask & GL_POLYGON_BIT) { struct gl_polygon_attrib *attr; attr = MALLOC_STRUCT( gl_polygon_attrib ); MEMCPY( attr, &ctx->Polygon, sizeof(struct gl_polygon_attrib) ); - newnode = new_attrib_node( GL_POLYGON_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_POLYGON_BIT, attr); } if (mask & GL_POLYGON_STIPPLE_BIT) { GLuint *stipple; stipple = (GLuint *) MALLOC( 32*sizeof(GLuint) ); MEMCPY( stipple, ctx->PolygonStipple, 32*sizeof(GLuint) ); - newnode = new_attrib_node( GL_POLYGON_STIPPLE_BIT ); - newnode->data = stipple; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_POLYGON_STIPPLE_BIT, stipple); } if (mask & GL_SCISSOR_BIT) { struct gl_scissor_attrib *attr; attr = MALLOC_STRUCT( gl_scissor_attrib ); MEMCPY( attr, &ctx->Scissor, sizeof(struct gl_scissor_attrib) ); - newnode = new_attrib_node( GL_SCISSOR_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_SCISSOR_BIT, attr); } if (mask & GL_STENCIL_BUFFER_BIT) { struct gl_stencil_attrib *attr; attr = MALLOC_STRUCT( gl_stencil_attrib ); MEMCPY( attr, &ctx->Stencil, sizeof(struct gl_stencil_attrib) ); - newnode = new_attrib_node( GL_STENCIL_BUFFER_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_STENCIL_BUFFER_BIT, attr); } if (mask & GL_TEXTURE_BIT) { @@ -494,30 +449,21 @@ _mesa_PushAttrib(GLbitfield mask) _mesa_unlock_context_textures(ctx); - newnode = new_attrib_node( GL_TEXTURE_BIT ); - newnode->data = texstate; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_TEXTURE_BIT, texstate); } if (mask & GL_TRANSFORM_BIT) { struct gl_transform_attrib *attr; attr = MALLOC_STRUCT( gl_transform_attrib ); MEMCPY( attr, &ctx->Transform, sizeof(struct gl_transform_attrib) ); - newnode = new_attrib_node( GL_TRANSFORM_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_TRANSFORM_BIT, attr); } if (mask & GL_VIEWPORT_BIT) { struct gl_viewport_attrib *attr; attr = MALLOC_STRUCT( gl_viewport_attrib ); MEMCPY( attr, &ctx->Viewport, sizeof(struct gl_viewport_attrib) ); - newnode = new_attrib_node( GL_VIEWPORT_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_VIEWPORT_BIT, attr); } /* GL_ARB_multisample */ @@ -525,10 +471,7 @@ _mesa_PushAttrib(GLbitfield mask) struct gl_multisample_attrib *attr; attr = MALLOC_STRUCT( gl_multisample_attrib ); MEMCPY( attr, &ctx->Multisample, sizeof(struct gl_multisample_attrib) ); - newnode = new_attrib_node( GL_MULTISAMPLE_BIT_ARB ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_MULTISAMPLE_BIT_ARB, attr); } end: @@ -1373,7 +1316,6 @@ copy_pixelstore(GLcontext *ctx, void GLAPIENTRY _mesa_PushClientAttrib(GLbitfield mask) { - struct gl_attrib_node *newnode; struct gl_attrib_node *head; GET_CURRENT_CONTEXT(ctx); @@ -1394,17 +1336,11 @@ _mesa_PushClientAttrib(GLbitfield mask) /* packing attribs */ attr = CALLOC_STRUCT( gl_pixelstore_attrib ); copy_pixelstore(ctx, attr, &ctx->Pack); - newnode = new_attrib_node( GL_CLIENT_PACK_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_CLIENT_PACK_BIT, attr); /* unpacking attribs */ attr = CALLOC_STRUCT( gl_pixelstore_attrib ); copy_pixelstore(ctx, attr, &ctx->Unpack); - newnode = new_attrib_node( GL_CLIENT_UNPACK_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_CLIENT_UNPACK_BIT, attr); } if (mask & GL_CLIENT_VERTEX_ARRAY_BIT) { @@ -1425,10 +1361,8 @@ _mesa_PushClientAttrib(GLbitfield mask) attr->ArrayObj = obj; - newnode = new_attrib_node( GL_CLIENT_VERTEX_ARRAY_BIT ); - newnode->data = attr; - newnode->next = head; - head = newnode; + save_attrib_data(&head, GL_CLIENT_VERTEX_ARRAY_BIT, attr); + /* bump reference counts on buffer objects */ adjust_buffer_object_ref_counts(ctx->Array.ArrayObj, 1); } diff --git a/src/mesa/main/bufferobj.c b/src/mesa/main/bufferobj.c index d640f5358e5..f96185a4b5d 100644 --- a/src/mesa/main/bufferobj.c +++ b/src/mesa/main/bufferobj.c @@ -156,7 +156,7 @@ buffer_object_subdata_range_good( GLcontext * ctx, GLenum target, _mesa_error(ctx, GL_INVALID_ENUM, "%s(target)", caller); return NULL; } - if (bufObj->Name == 0) { + if (!_mesa_is_bufferobj(bufObj)) { _mesa_error(ctx, GL_INVALID_OPERATION, "%s", caller); return NULL; } @@ -165,7 +165,7 @@ buffer_object_subdata_range_good( GLcontext * ctx, GLenum target, "%s(size + offset > buffer size)", caller); return NULL; } - if (bufObj->Pointer) { + if (_mesa_bufferobj_mapped(bufObj)) { /* Buffer is currently mapped */ _mesa_error(ctx, GL_INVALID_OPERATION, "%s", caller); return NULL; @@ -423,7 +423,7 @@ _mesa_buffer_map( GLcontext *ctx, GLenum target, GLenum access, (void) target; (void) access; /* Just return a direct pointer to the data */ - if (bufObj->Pointer) { + if (_mesa_bufferobj_mapped(bufObj)) { /* already mapped! */ return NULL; } @@ -445,7 +445,7 @@ _mesa_buffer_map_range( GLcontext *ctx, GLenum target, GLintptr offset, (void) target; (void) access; (void) length; - assert(!bufObj->Pointer); + assert(!_mesa_bufferobj_mapped(bufObj)); /* Just return a direct pointer to the data */ return bufObj->Data + offset; } @@ -502,8 +502,8 @@ _mesa_copy_buffer_subdata(GLcontext *ctx, GLubyte *srcPtr, *dstPtr; /* buffer should not already be mapped */ - assert(!src->Pointer); - assert(!dst->Pointer); + assert(!_mesa_bufferobj_mapped(src)); + assert(!_mesa_bufferobj_mapped(dst)); srcPtr = (GLubyte *) ctx->Driver.MapBuffer(ctx, GL_COPY_READ_BUFFER, GL_READ_ONLY, src); @@ -663,7 +663,7 @@ _mesa_validate_pbo_access(GLuint dimensions, GLvoid *start, *end; const GLubyte *sizeAddr; /* buffer size, cast to a pointer */ - ASSERT(pack->BufferObj->Name != 0); + ASSERT(_mesa_is_bufferobj(pack->BufferObj)); if (pack->BufferObj->Size == 0) /* no buffer! */ @@ -709,7 +709,7 @@ _mesa_map_bitmap_pbo(GLcontext *ctx, { const GLubyte *buf; - if (unpack->BufferObj->Name) { + if (_mesa_is_bufferobj(unpack->BufferObj)) { /* unpack from PBO */ buf = (GLubyte *) ctx->Driver.MapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, GL_READ_ONLY_ARB, @@ -736,7 +736,7 @@ void _mesa_unmap_bitmap_pbo(GLcontext *ctx, const struct gl_pixelstore_attrib *unpack) { - if (unpack->BufferObj->Name) { + if (_mesa_is_bufferobj(unpack->BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, unpack->BufferObj); } @@ -753,7 +753,7 @@ _mesa_map_drawpix_pbo(GLcontext *ctx, { const GLvoid *buf; - if (unpack->BufferObj->Name) { + if (_mesa_is_bufferobj(unpack->BufferObj)) { /* unpack from PBO */ buf = (GLubyte *) ctx->Driver.MapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, GL_READ_ONLY_ARB, @@ -779,7 +779,7 @@ void _mesa_unmap_drawpix_pbo(GLcontext *ctx, const struct gl_pixelstore_attrib *unpack) { - if (unpack->BufferObj->Name) { + if (_mesa_is_bufferobj(unpack->BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, unpack->BufferObj); } @@ -798,7 +798,7 @@ _mesa_map_readpix_pbo(GLcontext *ctx, { void *buf; - if (pack->BufferObj->Name) { + if (_mesa_is_bufferobj(pack->BufferObj)) { /* pack into PBO */ buf = (GLubyte *) ctx->Driver.MapBuffer(ctx, GL_PIXEL_PACK_BUFFER_EXT, GL_WRITE_ONLY_ARB, @@ -824,7 +824,7 @@ void _mesa_unmap_readpix_pbo(GLcontext *ctx, const struct gl_pixelstore_attrib *pack) { - if (pack->BufferObj->Name) { + if (_mesa_is_bufferobj(pack->BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_PACK_BUFFER_EXT, pack->BufferObj); } } @@ -932,7 +932,7 @@ _mesa_DeleteBuffersARB(GLsizei n, const GLuint *ids) ASSERT(bufObj->Name == ids[i]); - if (bufObj->Pointer) { + if (_mesa_bufferobj_mapped(bufObj)) { /* if mapped, unmap it now */ ctx->Driver.UnmapBuffer(ctx, 0, bufObj); bufObj->AccessFlags = DEFAULT_ACCESS; @@ -1086,12 +1086,12 @@ _mesa_BufferDataARB(GLenum target, GLsizeiptrARB size, _mesa_error(ctx, GL_INVALID_ENUM, "glBufferDataARB(target)" ); return; } - if (bufObj->Name == 0) { + if (!_mesa_is_bufferobj(bufObj)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glBufferDataARB(buffer 0)" ); return; } - if (bufObj->Pointer) { + if (_mesa_bufferobj_mapped(bufObj)) { /* Unmap the existing buffer. We'll replace it now. Not an error. */ ctx->Driver.UnmapBuffer(ctx, target, bufObj); bufObj->AccessFlags = DEFAULT_ACCESS; @@ -1187,18 +1187,18 @@ _mesa_MapBufferARB(GLenum target, GLenum access) _mesa_error(ctx, GL_INVALID_ENUM, "glMapBufferARB(target)" ); return NULL; } - if (bufObj->Name == 0) { + if (!_mesa_is_bufferobj(bufObj)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glMapBufferARB(buffer 0)" ); return NULL; } - if (bufObj->Pointer) { + if (_mesa_bufferobj_mapped(bufObj)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glMapBufferARB(already mapped)"); return NULL; } ASSERT(ctx->Driver.MapBuffer); bufObj->Pointer = ctx->Driver.MapBuffer( ctx, target, access, bufObj ); - if (!bufObj->Pointer) { + if (!_mesa_bufferobj_mapped(bufObj)) { _mesa_error(ctx, GL_OUT_OF_MEMORY, "glMapBufferARB(access)"); } @@ -1248,11 +1248,11 @@ _mesa_UnmapBufferARB(GLenum target) _mesa_error(ctx, GL_INVALID_ENUM, "glUnmapBufferARB(target)" ); return GL_FALSE; } - if (bufObj->Name == 0) { + if (!_mesa_is_bufferobj(bufObj)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glUnmapBufferARB" ); return GL_FALSE; } - if (!bufObj->Pointer) { + if (!_mesa_bufferobj_mapped(bufObj)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glUnmapBufferARB"); return GL_FALSE; } @@ -1315,7 +1315,7 @@ _mesa_GetBufferParameterivARB(GLenum target, GLenum pname, GLint *params) _mesa_error(ctx, GL_INVALID_ENUM, "GetBufferParameterivARB(target)" ); return; } - if (bufObj->Name == 0) { + if (!_mesa_is_bufferobj(bufObj)) { _mesa_error(ctx, GL_INVALID_OPERATION, "GetBufferParameterivARB" ); return; } @@ -1331,7 +1331,7 @@ _mesa_GetBufferParameterivARB(GLenum target, GLenum pname, GLint *params) *params = simplified_access_mode(bufObj->AccessFlags); break; case GL_BUFFER_MAPPED_ARB: - *params = (bufObj->Pointer != NULL); + *params = _mesa_bufferobj_mapped(bufObj); break; default: _mesa_error(ctx, GL_INVALID_ENUM, "glGetBufferParameterivARB(pname)"); @@ -1357,7 +1357,7 @@ _mesa_GetBufferPointervARB(GLenum target, GLenum pname, GLvoid **params) _mesa_error(ctx, GL_INVALID_ENUM, "glGetBufferPointervARB(target)" ); return; } - if (bufObj->Name == 0) { + if (!_mesa_is_bufferobj(bufObj)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glGetBufferPointervARB" ); return; } @@ -1376,26 +1376,26 @@ _mesa_CopyBufferSubData(GLenum readTarget, GLenum writeTarget, ASSERT_OUTSIDE_BEGIN_END(ctx); src = get_buffer(ctx, readTarget); - if (!src || src->Name == 0) { + if (!src || !_mesa_is_bufferobj(src)) { _mesa_error(ctx, GL_INVALID_ENUM, "glCopyBuffserSubData(readTarget = 0x%x)", readTarget); return; } dst = get_buffer(ctx, writeTarget); - if (!dst || dst->Name == 0) { + if (!dst || !_mesa_is_bufferobj(dst)) { _mesa_error(ctx, GL_INVALID_ENUM, "glCopyBuffserSubData(writeTarget = 0x%x)", writeTarget); return; } - if (src->Pointer) { + if (_mesa_bufferobj_mapped(src)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glCopyBuffserSubData(readBuffer is mapped)"); return; } - if (dst->Pointer) { + if (_mesa_bufferobj_mapped(dst)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glCopyBuffserSubData(writeBuffer is mapped)"); return; @@ -1499,7 +1499,7 @@ _mesa_MapBufferRange(GLenum target, GLintptr offset, GLsizeiptr length, } bufObj = get_buffer(ctx, target); - if (!bufObj || bufObj->Name == 0) { + if (!bufObj || !_mesa_is_bufferobj(bufObj)) { _mesa_error(ctx, GL_INVALID_ENUM, "glMapBufferRange(target = 0x%x)", target); return NULL; @@ -1511,7 +1511,7 @@ _mesa_MapBufferRange(GLenum target, GLintptr offset, GLsizeiptr length, return NULL; } - if (bufObj->Pointer) { + if (_mesa_bufferobj_mapped(bufObj)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glMapBufferRange(buffer already mapped)"); return NULL; @@ -1564,13 +1564,13 @@ _mesa_FlushMappedBufferRange(GLenum target, GLintptr offset, GLsizeiptr length) return; } - if (bufObj->Name == 0) { + if (!_mesa_is_bufferobj(bufObj)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glMapBufferRange(current buffer is 0)"); return; } - if (!bufObj->Pointer) { + if (!_mesa_bufferobj_mapped(bufObj)) { /* buffer is not mapped */ _mesa_error(ctx, GL_INVALID_OPERATION, "glMapBufferRange(buffer is not mapped)"); diff --git a/src/mesa/main/bufferobj.h b/src/mesa/main/bufferobj.h index ef59ff83c84..decb44a65e6 100644 --- a/src/mesa/main/bufferobj.h +++ b/src/mesa/main/bufferobj.h @@ -36,6 +36,26 @@ * Internal functions */ + +/** Is the given buffer object currently mapped? */ +static INLINE GLboolean +_mesa_bufferobj_mapped(const struct gl_buffer_object *obj) +{ + return obj->Pointer != NULL; +} + +/** + * Is the given buffer object a user-created buffer object? + * Mesa uses default buffer objects in several places. Default buffers + * always have Name==0. User created buffers have Name!=0. + */ +static INLINE GLboolean +_mesa_is_bufferobj(const struct gl_buffer_object *obj) +{ + return obj->Name != 0; +} + + extern void _mesa_init_buffer_objects( GLcontext *ctx ); diff --git a/src/mesa/main/colortab.c b/src/mesa/main/colortab.c index bd9cf438b4f..36304065eb7 100644 --- a/src/mesa/main/colortab.c +++ b/src/mesa/main/colortab.c @@ -179,7 +179,7 @@ store_colortable_entries(GLcontext *ctx, struct gl_color_table *table, GLfloat bScale, GLfloat bBias, GLfloat aScale, GLfloat aBias) { - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { /* Get/unpack the color table data from a PBO */ GLubyte *buf; if (!_mesa_validate_pbo_access(1, &ctx->Unpack, count, 1, 1, @@ -279,7 +279,7 @@ store_colortable_entries(GLcontext *ctx, struct gl_color_table *table, } } - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, ctx->Unpack.BufferObj); } @@ -696,7 +696,7 @@ _mesa_GetColorTable( GLenum target, GLenum format, return; } - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { /* pack color table into PBO */ GLubyte *buf; if (!_mesa_validate_pbo_access(1, &ctx->Pack, table->Size, 1, 1, @@ -720,7 +720,7 @@ _mesa_GetColorTable( GLenum target, GLenum format, _mesa_pack_rgba_span_float(ctx, table->Size, rgba, format, type, data, &ctx->Pack, 0x0); - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_PACK_BUFFER_EXT, ctx->Pack.BufferObj); } diff --git a/src/mesa/main/convolve.c b/src/mesa/main/convolve.c index 814c6a0a5a6..69dba72ed38 100644 --- a/src/mesa/main/convolve.c +++ b/src/mesa/main/convolve.c @@ -144,7 +144,7 @@ _mesa_ConvolutionFilter1D(GLenum target, GLenum internalFormat, GLsizei width, G ctx->Convolution1D.Width = width; ctx->Convolution1D.Height = 1; - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { /* unpack filter from PBO */ GLubyte *buf; if (!_mesa_validate_pbo_access(1, &ctx->Unpack, width, 1, 1, @@ -173,7 +173,7 @@ _mesa_ConvolutionFilter1D(GLenum target, GLenum internalFormat, GLsizei width, G format, type, image, &ctx->Unpack, 0); /* transferOps */ - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, ctx->Unpack.BufferObj); } @@ -242,7 +242,7 @@ _mesa_ConvolutionFilter2D(GLenum target, GLenum internalFormat, GLsizei width, G ctx->Convolution2D.Width = width; ctx->Convolution2D.Height = height; - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { /* unpack filter from PBO */ GLubyte *buf; if (!_mesa_validate_pbo_access(2, &ctx->Unpack, width, height, 1, @@ -276,7 +276,7 @@ _mesa_ConvolutionFilter2D(GLenum target, GLenum internalFormat, GLsizei width, G 0); /* transferOps */ } - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, ctx->Unpack.BufferObj); } @@ -598,7 +598,7 @@ _mesa_GetConvolutionFilter(GLenum target, GLenum format, GLenum type, return; } - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { /* Pack the filter into a PBO */ GLubyte *buf; if (!_mesa_validate_pbo_access(2, &ctx->Pack, @@ -629,7 +629,7 @@ _mesa_GetConvolutionFilter(GLenum target, GLenum format, GLenum type, format, type, dst, &ctx->Pack, 0x0); } - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_PACK_BUFFER_EXT, ctx->Pack.BufferObj); } @@ -802,7 +802,7 @@ _mesa_GetSeparableFilter(GLenum target, GLenum format, GLenum type, filter = &ctx->Separable2D; - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { /* Pack filter into PBO */ GLubyte *buf; if (!_mesa_validate_pbo_access(1, &ctx->Pack, filter->Width, 1, 1, @@ -850,7 +850,7 @@ _mesa_GetSeparableFilter(GLenum target, GLenum format, GLenum type, (void) span; /* unused at this time */ - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { /* Pack filter into PBO */ ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, ctx->Unpack.BufferObj); @@ -905,7 +905,7 @@ _mesa_SeparableFilter2D(GLenum target, GLenum internalFormat, GLsizei width, GLs ctx->Separable2D.Width = width; ctx->Separable2D.Height = height; - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { /* unpack filter from PBO */ GLubyte *buf; if (!_mesa_validate_pbo_access(1, &ctx->Unpack, width, 1, 1, @@ -971,7 +971,7 @@ _mesa_SeparableFilter2D(GLenum target, GLenum internalFormat, GLsizei width, GLs ctx->Pixel.ConvolutionFilterBias[2][3]); } - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, ctx->Unpack.BufferObj); } diff --git a/src/mesa/main/debug.c b/src/mesa/main/debug.c index 1c8c44fcb96..8492c8561d4 100644 --- a/src/mesa/main/debug.c +++ b/src/mesa/main/debug.c @@ -33,6 +33,7 @@ #include "get.h" #include "pixelstore.h" #include "readpix.h" +#include "texgetimage.h" #include "texobj.h" #include "texformat.h" @@ -234,11 +235,11 @@ _mesa_init_debug( GLcontext *ctx ) */ static void write_ppm(const char *filename, const GLubyte *buffer, int width, int height, - int comps, int rcomp, int gcomp, int bcomp) + int comps, int rcomp, int gcomp, int bcomp, GLboolean invert) { FILE *f = fopen( filename, "w" ); if (f) { - int i, x, y; + int x, y; const GLubyte *ptr = buffer; fprintf(f,"P6\n"); fprintf(f,"# ppm-file created by osdemo.c\n"); @@ -246,10 +247,11 @@ write_ppm(const char *filename, const GLubyte *buffer, int width, int height, fprintf(f,"255\n"); fclose(f); f = fopen( filename, "ab" ); /* reopen in binary append mode */ - for (y=height-1; y>=0; y--) { - for (x=0; x<width; x++) { - i = (y*width + x) * comps; - fputc(ptr[i+rcomp], f); /* write red */ + for (y=0; y < height; y++) { + for (x = 0; x < width; x++) { + int yy = invert ? (height - 1 - y) : y; + int i = (yy * width + x) * comps; + fputc(ptr[i+rcomp], f); /* write red */ fputc(ptr[i+gcomp], f); /* write green */ fputc(ptr[i+bcomp], f); /* write blue */ } @@ -263,47 +265,34 @@ write_ppm(const char *filename, const GLubyte *buffer, int width, int height, * Write level[0] image to a ppm file. */ static void -write_texture_image(struct gl_texture_object *texObj) +write_texture_image(struct gl_texture_object *texObj, GLuint face, GLuint level) { - const struct gl_texture_image *img = texObj->Image[0][0]; - if (img && img->Data) { + struct gl_texture_image *img = texObj->Image[face][level]; + if (img) { + GET_CURRENT_CONTEXT(ctx); + struct gl_pixelstore_attrib store; + GLubyte *buffer; char s[100]; + buffer = (GLubyte *) _mesa_malloc(img->Width * img->Height + * img->Depth * 4); + + store = ctx->Pack; /* save */ + ctx->Pack = ctx->DefaultPacking; + + ctx->Driver.GetTexImage(ctx, texObj->Target, level, + GL_RGBA, GL_UNSIGNED_BYTE, + buffer, texObj, img); + /* make filename */ - sprintf(s, "/tmp/teximage%u.ppm", texObj->Name); - - switch (img->TexFormat->MesaFormat) { - case MESA_FORMAT_RGBA8888: - write_ppm(s, img->Data, img->Width, img->Height, 4, 3, 2, 1); - break; - case MESA_FORMAT_ARGB8888: - write_ppm(s, img->Data, img->Width, img->Height, 4, 2, 1, 0); - break; - case MESA_FORMAT_RGB888: - write_ppm(s, img->Data, img->Width, img->Height, 3, 2, 1, 0); - break; - case MESA_FORMAT_RGB565: - { - GLubyte *buf2 = (GLubyte *) _mesa_malloc(img->Width * img->Height * 3); - GLuint i; - for (i = 0; i < img->Width * img->Height; i++) { - GLint r, g, b; - GLushort s = ((GLushort *) img->Data)[i]; - r = UBYTE_TO_CHAN( ((s >> 8) & 0xf8) | ((s >> 13) & 0x7) ); - g = UBYTE_TO_CHAN( ((s >> 3) & 0xfc) | ((s >> 9) & 0x3) ); - b = UBYTE_TO_CHAN( ((s << 3) & 0xf8) | ((s >> 2) & 0x7) ); - buf2[i*3+1] = r; - buf2[i*3+2] = g; - buf2[i*3+3] = b; - } - write_ppm(s, buf2, img->Width, img->Height, 3, 2, 1, 0); - _mesa_free(buf2); - } - break; - default: - printf("XXXX unsupported mesa tex format %d in %s\n", - img->TexFormat->MesaFormat, __FUNCTION__); - } + _mesa_sprintf(s, "/tmp/teximage%u.ppm", texObj->Name); + + _mesa_printf(" Writing image level %u to %s\n", level, s); + write_ppm(s, buffer, img->Width, img->Height, 4, 0, 1, 2, GL_FALSE); + + ctx->Pack = store; /* restore */ + + _mesa_free(buffer); } } @@ -316,17 +305,21 @@ dump_texture_cb(GLuint id, void *data, void *userData) { struct gl_texture_object *texObj = (struct gl_texture_object *) data; int i; + GLboolean written = GL_FALSE; (void) userData; - printf("Texture %u\n", texObj->Name); - printf(" Target 0x%x\n", texObj->Target); + _mesa_printf("Texture %u\n", texObj->Name); + _mesa_printf(" Target 0x%x\n", texObj->Target); for (i = 0; i < MAX_TEXTURE_LEVELS; i++) { struct gl_texture_image *texImg = texObj->Image[0][i]; if (texImg) { - printf(" Image %u: %d x %d x %d at %p\n", i, - texImg->Width, texImg->Height, texImg->Depth, texImg->Data); - if (DumpImages && i == 0) { - write_texture_image(texObj); + _mesa_printf(" Image %u: %d x %d x %d, format %u at %p\n", i, + texImg->Width, texImg->Height, texImg->Depth, + texImg->TexFormat->MesaFormat, texImg->Data); + if (DumpImages && !written) { + GLuint face = 0; + write_texture_image(texObj, face, i); + written = GL_TRUE; } } } @@ -368,7 +361,7 @@ _mesa_dump_color_buffer(const char *filename) ctx->DrawBuffer->_ColorDrawBuffers[0], ctx->DrawBuffer->ColorDrawBuffer[0]); _mesa_printf("Writing %d x %d color buffer to %s\n", w, h, filename); - write_ppm(filename, buf, w, h, 4, 0, 1, 2); + write_ppm(filename, buf, w, h, 4, 0, 1, 2, GL_TRUE); _mesa_PopClientAttrib(); @@ -403,7 +396,7 @@ _mesa_dump_depth_buffer(const char *filename) } _mesa_printf("Writing %d x %d depth buffer to %s\n", w, h, filename); - write_ppm(filename, buf2, w, h, 3, 0, 1, 2); + write_ppm(filename, buf2, w, h, 3, 0, 1, 2, GL_TRUE); _mesa_PopClientAttrib(); @@ -438,7 +431,7 @@ _mesa_dump_stencil_buffer(const char *filename) } _mesa_printf("Writing %d x %d stencil buffer to %s\n", w, h, filename); - write_ppm(filename, buf2, w, h, 3, 0, 1, 2); + write_ppm(filename, buf2, w, h, 3, 0, 1, 2, GL_TRUE); _mesa_PopClientAttrib(); diff --git a/src/mesa/main/dlist.c b/src/mesa/main/dlist.c index 49f202daa1c..4133f15dc72 100644 --- a/src/mesa/main/dlist.c +++ b/src/mesa/main/dlist.c @@ -690,7 +690,7 @@ unpack_image(GLuint dimensions, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const GLvoid * pixels, const struct gl_pixelstore_attrib *unpack) { - if (unpack->BufferObj->Name == 0) { + if (!_mesa_is_bufferobj(unpack->BufferObj)) { /* no PBO */ return _mesa_unpack_image(dimensions, width, height, depth, format, type, pixels, unpack); diff --git a/src/mesa/main/drawpix.c b/src/mesa/main/drawpix.c index 6682b5e725c..6d31f32443a 100644 --- a/src/mesa/main/drawpix.c +++ b/src/mesa/main/drawpix.c @@ -27,6 +27,7 @@ #include "bufferobj.h" #include "context.h" #include "drawpix.h" +#include "enums.h" #include "feedback.h" #include "framebuffer.h" #include "image.h" @@ -34,6 +35,18 @@ #include "state.h" + +/** + * If a fragment program is enabled, check that it's valid. + * \return GL_TRUE if valid, GL_FALSE otherwise + */ +static GLboolean +valid_fragment_program(GLcontext *ctx) +{ + return !(ctx->FragmentProgram.Enabled && !ctx->FragmentProgram._Enabled); +} + + #if _HAVE_FULL_GL /* @@ -51,29 +64,34 @@ _mesa_DrawPixels( GLsizei width, GLsizei height, return; } + /* We're not using the current vertex program, and the driver may install + * it's own. + */ + _mesa_set_vp_override(ctx, GL_TRUE); + if (ctx->NewState) { _mesa_update_state(ctx); } - if (ctx->FragmentProgram.Enabled && !ctx->FragmentProgram._Enabled) { + if (!valid_fragment_program(ctx)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawPixels (invalid fragment program)"); - return; + goto end; } if (_mesa_error_check_format_type(ctx, format, type, GL_TRUE)) { - /* found an error */ - return; + /* the error was already recorded */ + goto end; } if (ctx->DrawBuffer->_Status != GL_FRAMEBUFFER_COMPLETE_EXT) { _mesa_error(ctx, GL_INVALID_FRAMEBUFFER_OPERATION_EXT, "glDrawPixels(incomplete framebuffer)" ); - return; + goto end; } if (!ctx->Current.RasterPosValid) { - return; + goto end; /* no-op, not an error */ } if (ctx->RenderMode == GL_RENDER) { @@ -88,13 +106,13 @@ _mesa_DrawPixels( GLsizei width, GLsizei height, format, type, pixels)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawPixels(invalid PBO access)"); - return; + goto end; } - if (ctx->Unpack.BufferObj->Pointer) { + if (_mesa_bufferobj_mapped(ctx->Unpack.BufferObj)) { /* buffer is mapped - that's an error */ _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawPixels(PBO is mapped)"); - return; + goto end; } } @@ -116,6 +134,9 @@ _mesa_DrawPixels( GLsizei width, GLsizei height, ASSERT(ctx->RenderMode == GL_SELECT); /* Do nothing. See OpenGL Spec, Appendix B, Corollary 6. */ } + +end: + _mesa_set_vp_override(ctx, GL_FALSE); } @@ -126,37 +147,48 @@ _mesa_CopyPixels( GLint srcx, GLint srcy, GLsizei width, GLsizei height, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (width < 0 || height < 0) { + _mesa_error(ctx, GL_INVALID_VALUE, "glCopyPixels(width or height < 0)"); + return; + } + + if (type != GL_COLOR && type != GL_DEPTH && type != GL_STENCIL) { + _mesa_error(ctx, GL_INVALID_ENUM, "glCopyPixels(type=%s)", + _mesa_lookup_enum_by_nr(type)); + return; + } + + /* We're not using the current vertex program, and the driver may install + * it's own. + */ + _mesa_set_vp_override(ctx, GL_TRUE); + if (ctx->NewState) { _mesa_update_state(ctx); } - if (ctx->FragmentProgram.Enabled && !ctx->FragmentProgram._Enabled) { + if (!valid_fragment_program(ctx)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glCopyPixels (invalid fragment program)"); - return; - } - - if (width < 0 || height < 0) { - _mesa_error(ctx, GL_INVALID_VALUE, "glCopyPixels(width or height < 0)"); - return; + goto end; } if (ctx->DrawBuffer->_Status != GL_FRAMEBUFFER_COMPLETE_EXT || ctx->ReadBuffer->_Status != GL_FRAMEBUFFER_COMPLETE_EXT) { _mesa_error(ctx, GL_INVALID_FRAMEBUFFER_OPERATION_EXT, "glCopyPixels(incomplete framebuffer)" ); - return; + goto end; } if (!_mesa_source_buffer_exists(ctx, type) || !_mesa_dest_buffer_exists(ctx, type)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glCopyPixels(missing source or dest buffer)"); - return; + goto end; } if (!ctx->Current.RasterPosValid || width ==0 || height == 0) { - return; + goto end; /* no-op, not an error */ } if (ctx->RenderMode == GL_RENDER) { @@ -181,6 +213,9 @@ _mesa_CopyPixels( GLint srcx, GLint srcy, GLsizei width, GLsizei height, ASSERT(ctx->RenderMode == GL_SELECT); /* Do nothing. See OpenGL Spec, Appendix B, Corollary 6. */ } + +end: + _mesa_set_vp_override(ctx, GL_FALSE); } #endif /* _HAVE_FULL_GL */ @@ -208,7 +243,7 @@ _mesa_Bitmap( GLsizei width, GLsizei height, _mesa_update_state(ctx); } - if (ctx->FragmentProgram.Enabled && !ctx->FragmentProgram._Enabled) { + if (!valid_fragment_program(ctx)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glBitmap (invalid fragment program)"); return; @@ -235,7 +270,7 @@ _mesa_Bitmap( GLsizei width, GLsizei height, "glBitmap(invalid PBO access)"); return; } - if (ctx->Unpack.BufferObj->Pointer) { + if (_mesa_bufferobj_mapped(ctx->Unpack.BufferObj)) { /* buffer is mapped - that's an error */ _mesa_error(ctx, GL_INVALID_OPERATION, "glBitmap(PBO is mapped)"); return; @@ -264,68 +299,3 @@ _mesa_Bitmap( GLsizei width, GLsizei height, ctx->Current.RasterPos[0] += xmove; ctx->Current.RasterPos[1] += ymove; } - - - -#if 0 /* experimental */ -/* - * Execute glDrawDepthPixelsMESA(). This function accepts both a color - * image and depth (Z) image. Rasterization produces fragments with - * color and Z taken from these images. This function is intended for - * Z-compositing. Normally, this operation requires two glDrawPixels - * calls with stencil testing. - */ -void GLAPIENTRY -_mesa_DrawDepthPixelsMESA( GLsizei width, GLsizei height, - GLenum colorFormat, GLenum colorType, - const GLvoid *colors, - GLenum depthType, const GLvoid *depths ) -{ - GET_CURRENT_CONTEXT(ctx); - ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); - - if (width < 0 || height < 0) { - _mesa_error( ctx, GL_INVALID_VALUE, - "glDrawDepthPixelsMESA(width or height < 0" ); - return; - } - - if (!ctx->Current.RasterPosValid) { - return; - } - - if (ctx->NewState) { - _mesa_update_state(ctx); - } - - if (ctx->DrawBuffer->_Status != GL_FRAMEBUFFER_COMPLETE_EXT) { - _mesa_error(ctx, GL_INVALID_FRAMEBUFFER_OPERATION_EXT, - "glDrawDepthPixelsMESA(incomplete framebuffer)"); - return; - } - - if (ctx->RenderMode == GL_RENDER) { - /* Round, to satisfy conformance tests (matches SGI's OpenGL) */ - GLint x = IROUND(ctx->Current.RasterPos[0]); - GLint y = IROUND(ctx->Current.RasterPos[1]); - ctx->Driver.DrawDepthPixelsMESA(ctx, x, y, width, height, - colorFormat, colorType, colors, - depthType, depths, &ctx->Unpack); - } - else if (ctx->RenderMode == GL_FEEDBACK) { - /* Feedback the current raster pos info */ - FLUSH_CURRENT( ctx, 0 ); - _mesa_feedback_token( ctx, (GLfloat) (GLint) GL_DRAW_PIXEL_TOKEN ); - _mesa_feedback_vertex( ctx, - ctx->Current.RasterPos, - ctx->Current.RasterColor, - ctx->Current.RasterIndex, - ctx->Current.RasterTexCoords[0] ); - } - else { - ASSERT(ctx->RenderMode == GL_SELECT); - /* Do nothing. See OpenGL Spec, Appendix B, Corollary 6. */ - } -} - -#endif diff --git a/src/mesa/main/extensions.c b/src/mesa/main/extensions.c index c60b58a4926..8870a20d0ef 100644 --- a/src/mesa/main/extensions.c +++ b/src/mesa/main/extensions.c @@ -78,6 +78,7 @@ static const struct { { OFF, "GL_ARB_texture_non_power_of_two", F(ARB_texture_non_power_of_two)}, { OFF, "GL_ARB_texture_rectangle", F(NV_texture_rectangle) }, { ON, "GL_ARB_transpose_matrix", F(ARB_transpose_matrix) }, + { OFF, "GL_ARB_vertex_array_bgra", F(EXT_vertex_array_bgra) }, { OFF, "GL_ARB_vertex_array_object", F(ARB_vertex_array_object) }, { ON, "GL_ARB_vertex_buffer_object", F(ARB_vertex_buffer_object) }, { OFF, "GL_ARB_vertex_program", F(ARB_vertex_program) }, @@ -413,6 +414,7 @@ _mesa_enable_2_0_extensions(GLcontext *ctx) ctx->Extensions.ARB_fragment_shader = GL_TRUE; #endif ctx->Extensions.ARB_point_sprite = GL_TRUE; + ctx->Extensions.EXT_blend_equation_separate = GL_TRUE; ctx->Extensions.ARB_texture_non_power_of_two = GL_TRUE; #if FEATURE_ARB_shader_objects ctx->Extensions.ARB_shader_objects = GL_TRUE; diff --git a/src/mesa/main/fbobject.c b/src/mesa/main/fbobject.c index 83301f1e621..ab91fbc4de9 100644 --- a/src/mesa/main/fbobject.c +++ b/src/mesa/main/fbobject.c @@ -2038,7 +2038,9 @@ _mesa_BlitFramebufferEXT(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, if (mask & GL_STENCIL_BUFFER_BIT) { struct gl_renderbuffer *readRb = readFb->_StencilBuffer; struct gl_renderbuffer *drawRb = drawFb->_StencilBuffer; - if (readRb->StencilBits != drawRb->StencilBits) { + if (!readRb || + !drawRb || + readRb->StencilBits != drawRb->StencilBits) { _mesa_error(ctx, GL_INVALID_OPERATION, "glBlitFramebufferEXT(stencil buffer size mismatch"); return; @@ -2048,7 +2050,9 @@ _mesa_BlitFramebufferEXT(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, if (mask & GL_DEPTH_BUFFER_BIT) { struct gl_renderbuffer *readRb = readFb->_DepthBuffer; struct gl_renderbuffer *drawRb = drawFb->_DepthBuffer; - if (readRb->DepthBits != drawRb->DepthBits) { + if (!readRb || + !drawRb || + readRb->DepthBits != drawRb->DepthBits) { _mesa_error(ctx, GL_INVALID_OPERATION, "glBlitFramebufferEXT(depth buffer size mismatch"); return; diff --git a/src/mesa/main/framebuffer.c b/src/mesa/main/framebuffer.c index 5a13c88a7ab..dc79b8ca61a 100644 --- a/src/mesa/main/framebuffer.c +++ b/src/mesa/main/framebuffer.c @@ -817,7 +817,7 @@ _mesa_update_framebuffer(GLcontext *ctx) /** * Check if the renderbuffer for a read operation (glReadPixels, glCopyPixels, - * glCopyTex[Sub]Image, etc. exists. + * glCopyTex[Sub]Image, etc) exists. * \param format a basic image format such as GL_RGB, GL_RGBA, GL_ALPHA, * GL_DEPTH_COMPONENT, etc. or GL_COLOR, GL_DEPTH, GL_STENCIL. * \return GL_TRUE if buffer exists, GL_FALSE otherwise @@ -825,8 +825,12 @@ _mesa_update_framebuffer(GLcontext *ctx) GLboolean _mesa_source_buffer_exists(GLcontext *ctx, GLenum format) { - const struct gl_renderbuffer_attachment *att - = ctx->ReadBuffer->Attachment; + const struct gl_renderbuffer_attachment *att = ctx->ReadBuffer->Attachment; + + /* If we don't know the framebuffer status, update it now */ + if (ctx->ReadBuffer->_Status == 0) { + _mesa_test_framebuffer_completeness(ctx, ctx->ReadBuffer); + } if (ctx->ReadBuffer->_Status != GL_FRAMEBUFFER_COMPLETE_EXT) { return GL_FALSE; @@ -850,10 +854,8 @@ _mesa_source_buffer_exists(GLcontext *ctx, GLenum format) if (ctx->ReadBuffer->_ColorReadBuffer == NULL) { return GL_FALSE; } - /* XXX enable this post 6.5 release: ASSERT(ctx->ReadBuffer->_ColorReadBuffer->RedBits > 0 || ctx->ReadBuffer->_ColorReadBuffer->IndexBits > 0); - */ break; case GL_DEPTH: case GL_DEPTH_COMPONENT: @@ -891,13 +893,17 @@ _mesa_source_buffer_exists(GLcontext *ctx, GLenum format) /** * As above, but for drawing operations. - * XXX code do some code merging w/ above function. + * XXX could do some code merging w/ above function. */ GLboolean _mesa_dest_buffer_exists(GLcontext *ctx, GLenum format) { - const struct gl_renderbuffer_attachment *att - = ctx->ReadBuffer->Attachment; + const struct gl_renderbuffer_attachment *att = ctx->DrawBuffer->Attachment; + + /* If we don't know the framebuffer status, update it now */ + if (ctx->DrawBuffer->_Status == 0) { + _mesa_test_framebuffer_completeness(ctx, ctx->DrawBuffer); + } if (ctx->DrawBuffer->_Status != GL_FRAMEBUFFER_COMPLETE_EXT) { return GL_FALSE; @@ -918,7 +924,7 @@ _mesa_dest_buffer_exists(GLcontext *ctx, GLenum format) case GL_BGRA: case GL_ABGR_EXT: case GL_COLOR_INDEX: - /* nothing special */ + /* Nothing special since GL_DRAW_BUFFER could be GL_NONE. */ /* Could assert that colorbuffer has RedBits > 0 */ break; case GL_DEPTH: @@ -945,7 +951,7 @@ _mesa_dest_buffer_exists(GLcontext *ctx, GLenum format) break; default: _mesa_problem(ctx, - "Unexpected format 0x%x in _mesa_source_buffer_exists", + "Unexpected format 0x%x in _mesa_dest_buffer_exists", format); return GL_FALSE; } diff --git a/src/mesa/main/getstring.c b/src/mesa/main/getstring.c index 41fd786d7d5..6599ed9698d 100644 --- a/src/mesa/main/getstring.c +++ b/src/mesa/main/getstring.c @@ -266,5 +266,6 @@ _mesa_GetError( void ) _mesa_debug(ctx, "glGetError <-- %s\n", _mesa_lookup_enum_by_nr(e)); ctx->ErrorValue = (GLenum) GL_NO_ERROR; + ctx->ErrorDebugCount = 0; return e; } diff --git a/src/mesa/main/histogram.c b/src/mesa/main/histogram.c index 5fee4fd0e34..726a50d3b1b 100644 --- a/src/mesa/main/histogram.c +++ b/src/mesa/main/histogram.c @@ -649,7 +649,7 @@ _mesa_GetMinmax(GLenum target, GLboolean reset, GLenum format, GLenum type, GLvo return; } - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { /* pack min/max values into a PBO */ GLubyte *buf; if (!_mesa_validate_pbo_access(1, &ctx->Pack, 2, 1, 1, @@ -687,7 +687,7 @@ _mesa_GetMinmax(GLenum target, GLboolean reset, GLenum format, GLenum type, GLvo format, type, values, &ctx->Pack, 0x0); } - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_PACK_BUFFER_EXT, ctx->Pack.BufferObj); } @@ -733,7 +733,7 @@ _mesa_GetHistogram(GLenum target, GLboolean reset, GLenum format, GLenum type, G return; } - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { /* pack min/max values into a PBO */ GLubyte *buf; if (!_mesa_validate_pbo_access(1, &ctx->Pack, ctx->Histogram.Width, 1, 1, @@ -761,7 +761,7 @@ _mesa_GetHistogram(GLenum target, GLboolean reset, GLenum format, GLenum type, G (CONST GLuint (*)[4]) ctx->Histogram.Count, format, type, values, &ctx->Pack); - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_PACK_BUFFER_EXT, ctx->Pack.BufferObj); } diff --git a/src/mesa/main/image.c b/src/mesa/main/image.c index 090e5eb3300..e0e8548d7cd 100644 --- a/src/mesa/main/image.c +++ b/src/mesa/main/image.c @@ -5335,3 +5335,181 @@ _mesa_clip_to_region(GLint xmin, GLint ymin, return GL_TRUE; } + + +/** + * Clip dst coords against Xmax (or Ymax). + */ +static INLINE void +clip_right_or_top(GLint *srcX0, GLint *srcX1, + GLint *dstX0, GLint *dstX1, + GLint maxValue) +{ + GLfloat t, bias; + + if (*dstX1 > maxValue) { + /* X1 outside right edge */ + ASSERT(*dstX0 < maxValue); /* X0 should be inside right edge */ + t = (GLfloat) (maxValue - *dstX0) / (GLfloat) (*dstX1 - *dstX0); + /* chop off [t, 1] part */ + ASSERT(t >= 0.0 && t <= 1.0); + *dstX1 = maxValue; + bias = (*srcX0 < *srcX1) ? 0.5 : -0.5; + *srcX1 = *srcX0 + (GLint) (t * (*srcX1 - *srcX0) + bias); + } + else if (*dstX0 > maxValue) { + /* X0 outside right edge */ + ASSERT(*dstX1 < maxValue); /* X1 should be inside right edge */ + t = (GLfloat) (maxValue - *dstX1) / (GLfloat) (*dstX0 - *dstX1); + /* chop off [t, 1] part */ + ASSERT(t >= 0.0 && t <= 1.0); + *dstX0 = maxValue; + bias = (*srcX0 < *srcX1) ? -0.5 : 0.5; + *srcX0 = *srcX1 + (GLint) (t * (*srcX0 - *srcX1) + bias); + } +} + + +/** + * Clip dst coords against Xmin (or Ymin). + */ +static INLINE void +clip_left_or_bottom(GLint *srcX0, GLint *srcX1, + GLint *dstX0, GLint *dstX1, + GLint minValue) +{ + GLfloat t, bias; + + if (*dstX0 < minValue) { + /* X0 outside left edge */ + ASSERT(*dstX1 > minValue); /* X1 should be inside left edge */ + t = (GLfloat) (minValue - *dstX0) / (GLfloat) (*dstX1 - *dstX0); + /* chop off [0, t] part */ + ASSERT(t >= 0.0 && t <= 1.0); + *dstX0 = minValue; + bias = (*srcX0 < *srcX1) ? 0.5 : -0.5; /* flipped??? */ + *srcX0 = *srcX0 + (GLint) (t * (*srcX1 - *srcX0) + bias); + } + else if (*dstX1 < minValue) { + /* X1 outside left edge */ + ASSERT(*dstX0 > minValue); /* X0 should be inside left edge */ + t = (GLfloat) (minValue - *dstX1) / (GLfloat) (*dstX0 - *dstX1); + /* chop off [0, t] part */ + ASSERT(t >= 0.0 && t <= 1.0); + *dstX1 = minValue; + bias = (*srcX0 < *srcX1) ? 0.5 : -0.5; + *srcX1 = *srcX1 + (GLint) (t * (*srcX0 - *srcX1) + bias); + } +} + + +/** + * Do clipping of blit src/dest rectangles. + * The dest rect is clipped against both the buffer bounds and scissor bounds. + * The src rect is just clipped against the buffer bounds. + * + * When either the src or dest rect is clipped, the other is also clipped + * proportionately! + * + * Note that X0 need not be less than X1 (same for Y) for either the source + * and dest rects. That makes the clipping a little trickier. + * + * \return GL_TRUE if anything is left to draw, GL_FALSE if totally clipped + */ +GLboolean +_mesa_clip_blit(GLcontext *ctx, + GLint *srcX0, GLint *srcY0, GLint *srcX1, GLint *srcY1, + GLint *dstX0, GLint *dstY0, GLint *dstX1, GLint *dstY1) +{ + const GLint srcXmin = 0; + const GLint srcXmax = ctx->ReadBuffer->Width; + const GLint srcYmin = 0; + const GLint srcYmax = ctx->ReadBuffer->Height; + + /* these include scissor bounds */ + const GLint dstXmin = ctx->DrawBuffer->_Xmin; + const GLint dstXmax = ctx->DrawBuffer->_Xmax; + const GLint dstYmin = ctx->DrawBuffer->_Ymin; + const GLint dstYmax = ctx->DrawBuffer->_Ymax; + + /* + printf("PreClipX: src: %d .. %d dst: %d .. %d\n", + *srcX0, *srcX1, *dstX0, *dstX1); + printf("PreClipY: src: %d .. %d dst: %d .. %d\n", + *srcY0, *srcY1, *dstY0, *dstY1); + */ + + /* trivial rejection tests */ + if (*dstX0 == *dstX1) + return GL_FALSE; /* no width */ + if (*dstX0 <= dstXmin && *dstX1 <= dstXmin) + return GL_FALSE; /* totally out (left) of bounds */ + if (*dstX0 >= dstXmax && *dstX1 >= dstXmax) + return GL_FALSE; /* totally out (right) of bounds */ + + if (*dstY0 == *dstY1) + return GL_FALSE; + if (*dstY0 <= dstYmin && *dstY1 <= dstYmin) + return GL_FALSE; + if (*dstY0 >= dstYmax && *dstY1 >= dstYmax) + return GL_FALSE; + + if (*srcX0 == *srcX1) + return GL_FALSE; + if (*srcX0 <= srcXmin && *srcX1 <= srcXmin) + return GL_FALSE; + if (*srcX0 >= srcXmax && *srcX1 >= srcXmax) + return GL_FALSE; + + if (*srcY0 == *srcY1) + return GL_FALSE; + if (*srcY0 <= srcYmin && *srcY1 <= srcYmin) + return GL_FALSE; + if (*srcY0 >= srcYmax && *srcY1 >= srcYmax) + return GL_FALSE; + + /* + * dest clip + */ + clip_right_or_top(srcX0, srcX1, dstX0, dstX1, dstXmax); + clip_right_or_top(srcY0, srcY1, dstY0, dstY1, dstYmax); + clip_left_or_bottom(srcX0, srcX1, dstX0, dstX1, dstXmin); + clip_left_or_bottom(srcY0, srcY1, dstY0, dstY1, dstYmin); + + /* + * src clip (just swap src/dst values from above) + */ + clip_right_or_top(dstX0, dstX1, srcX0, srcX1, srcXmax); + clip_right_or_top(dstY0, dstY1, srcY0, srcY1, srcYmax); + clip_left_or_bottom(dstX0, dstX1, srcX0, srcX1, srcXmin); + clip_left_or_bottom(dstY0, dstY1, srcY0, srcY1, srcYmin); + + /* + printf("PostClipX: src: %d .. %d dst: %d .. %d\n", + *srcX0, *srcX1, *dstX0, *dstX1); + printf("PostClipY: src: %d .. %d dst: %d .. %d\n", + *srcY0, *srcY1, *dstY0, *dstY1); + */ + + ASSERT(*dstX0 >= dstXmin); + ASSERT(*dstX0 <= dstXmax); + ASSERT(*dstX1 >= dstXmin); + ASSERT(*dstX1 <= dstXmax); + + ASSERT(*dstY0 >= dstYmin); + ASSERT(*dstY0 <= dstYmax); + ASSERT(*dstY1 >= dstYmin); + ASSERT(*dstY1 <= dstYmax); + + ASSERT(*srcX0 >= srcXmin); + ASSERT(*srcX0 <= srcXmax); + ASSERT(*srcX1 >= srcXmin); + ASSERT(*srcX1 <= srcXmax); + + ASSERT(*srcY0 >= srcYmin); + ASSERT(*srcY0 <= srcYmax); + ASSERT(*srcY1 >= srcYmin); + ASSERT(*srcY1 <= srcYmax); + + return GL_TRUE; +} diff --git a/src/mesa/main/image.h b/src/mesa/main/image.h index b26c27e5a8a..ee17accb80c 100644 --- a/src/mesa/main/image.h +++ b/src/mesa/main/image.h @@ -291,4 +291,10 @@ _mesa_clip_to_region(GLint xmin, GLint ymin, GLint *x, GLint *y, GLsizei *width, GLsizei *height ); +extern GLboolean +_mesa_clip_blit(GLcontext *ctx, + GLint *srcX0, GLint *srcY0, GLint *srcX1, GLint *srcY1, + GLint *dstX0, GLint *dstY0, GLint *dstX1, GLint *dstY1); + + #endif diff --git a/src/mesa/main/imports.c b/src/mesa/main/imports.c index 1722579e82c..6ffaddcde96 100644 --- a/src/mesa/main/imports.c +++ b/src/mesa/main/imports.c @@ -911,6 +911,20 @@ _mesa_strtod( const char *s, char **end ) return strtod(s, end); } +/** Compute simple checksum/hash for a string */ +unsigned int +_mesa_str_checksum(const char *str) +{ + /* This could probably be much better */ + unsigned int sum, i; + const char *c; + sum = i = 1; + for (c = str; *c; c++) + sum += *c * (i % 100); + return sum; +} + + /*@}*/ @@ -1021,23 +1035,59 @@ output_if_debug(const char *prefixString, const char *outputString, } } -static const char *error_string( GLenum error ); -static void flush_delayed_errors( GLcontext *ctx ) +/** + * Return string version of GL error code. + */ +static const char * +error_string( GLenum error ) +{ + switch (error) { + case GL_NO_ERROR: + return "GL_NO_ERROR"; + case GL_INVALID_VALUE: + return "GL_INVALID_VALUE"; + case GL_INVALID_ENUM: + return "GL_INVALID_ENUM"; + case GL_INVALID_OPERATION: + return "GL_INVALID_OPERATION"; + case GL_STACK_OVERFLOW: + return "GL_STACK_OVERFLOW"; + case GL_STACK_UNDERFLOW: + return "GL_STACK_UNDERFLOW"; + case GL_OUT_OF_MEMORY: + return "GL_OUT_OF_MEMORY"; + case GL_TABLE_TOO_LARGE: + return "GL_TABLE_TOO_LARGE"; + case GL_INVALID_FRAMEBUFFER_OPERATION_EXT: + return "GL_INVALID_FRAMEBUFFER_OPERATION"; + default: + return "unknown"; + } +} + + +/** + * When a new type of error is recorded, print a message describing + * previous errors which were accumulated. + */ +static void +flush_delayed_errors( GLcontext *ctx ) { - char s2[MAXSTRING]; + char s[MAXSTRING]; if (ctx->ErrorDebugCount) { - _mesa_snprintf(s2, MAXSTRING, "%d similar %s errors", + _mesa_snprintf(s, MAXSTRING, "%d similar %s errors", ctx->ErrorDebugCount, error_string(ctx->ErrorValue)); - output_if_debug("Mesa: ", s2, GL_TRUE); + output_if_debug("Mesa", s, GL_TRUE); ctx->ErrorDebugCount = 0; } } + /** * Report a warning (a recoverable error condition) to stderr if * either DEBUG is defined or the MESA_DEBUG env var is set. @@ -1083,31 +1133,6 @@ _mesa_problem( const GLcontext *ctx, const char *fmtString, ... ) fprintf(stderr, "Please report at bugzilla.freedesktop.org\n"); } -static const char *error_string( GLenum error ) -{ - switch (error) { - case GL_NO_ERROR: - return "GL_NO_ERROR"; - case GL_INVALID_VALUE: - return "GL_INVALID_VALUE"; - case GL_INVALID_ENUM: - return "GL_INVALID_ENUM"; - case GL_INVALID_OPERATION: - return "GL_INVALID_OPERATION"; - case GL_STACK_OVERFLOW: - return "GL_STACK_OVERFLOW"; - case GL_STACK_UNDERFLOW: - return "GL_STACK_UNDERFLOW"; - case GL_OUT_OF_MEMORY: - return "GL_OUT_OF_MEMORY"; - case GL_TABLE_TOO_LARGE: - return "GL_TABLE_TOO_LARGE"; - case GL_INVALID_FRAMEBUFFER_OPERATION_EXT: - return "GL_INVALID_FRAMEBUFFER_OPERATION"; - default: - return "unknown"; - } -} /** * Record an OpenGL state error. These usually occur when the user diff --git a/src/mesa/main/imports.h b/src/mesa/main/imports.h index 7b61e22e932..fb85f0862c6 100644 --- a/src/mesa/main/imports.h +++ b/src/mesa/main/imports.h @@ -586,6 +586,9 @@ _mesa_atoi( const char *s ); extern double _mesa_strtod( const char *s, char **end ); +extern unsigned int +_mesa_str_checksum(const char *str); + extern int _mesa_sprintf( char *str, const char *fmt, ... ); diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h index da01c58c2d8..6a60ad1ceee 100644 --- a/src/mesa/main/mtypes.h +++ b/src/mesa/main/mtypes.h @@ -83,6 +83,7 @@ /*@{*/ struct _mesa_HashTable; struct gl_attrib_node; +struct gl_meta_state; struct gl_pixelstore_attrib; struct gl_program_cache; struct gl_texture_format; @@ -2005,6 +2006,7 @@ struct gl_shader GLboolean Main; /**< shader defines main() */ GLboolean UnresolvedRefs; const GLchar *Source; /**< Source code string */ + GLuint SourceChecksum; /**< for debug/logging purposes */ struct gl_program *Program; /**< Post-compile assembly code */ GLchar *InfoLog; struct gl_sl_pragmas Pragmas; @@ -2982,6 +2984,8 @@ struct __GLcontextRec struct gl_buffer_object *CopyWriteBuffer; /**< GL_ARB_copy_buffer */ /*@}*/ + struct gl_meta_state *Meta; /**< for "meta" operations */ + #if FEATURE_EXT_framebuffer_object struct gl_renderbuffer *CurrentRenderbuffer; #endif diff --git a/src/mesa/main/pixel.c b/src/mesa/main/pixel.c index d9f3e476e81..25f55a422fa 100644 --- a/src/mesa/main/pixel.c +++ b/src/mesa/main/pixel.c @@ -158,7 +158,7 @@ _mesa_PixelMapfv( GLenum map, GLsizei mapsize, const GLfloat *values ) FLUSH_VERTICES(ctx, _NEW_PIXEL); - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { /* unpack pixelmap from PBO */ GLubyte *buf; /* Note, need to use DefaultPacking and Unpack's buffer object */ @@ -188,7 +188,7 @@ _mesa_PixelMapfv( GLenum map, GLsizei mapsize, const GLfloat *values ) store_pixelmap(ctx, map, mapsize, values); - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, ctx->Unpack.BufferObj); } @@ -217,7 +217,7 @@ _mesa_PixelMapuiv(GLenum map, GLsizei mapsize, const GLuint *values ) FLUSH_VERTICES(ctx, _NEW_PIXEL); - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { /* unpack pixelmap from PBO */ GLubyte *buf; /* Note, need to use DefaultPacking and Unpack's buffer object */ @@ -259,7 +259,7 @@ _mesa_PixelMapuiv(GLenum map, GLsizei mapsize, const GLuint *values ) } } - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, ctx->Unpack.BufferObj); } @@ -290,7 +290,7 @@ _mesa_PixelMapusv(GLenum map, GLsizei mapsize, const GLushort *values ) FLUSH_VERTICES(ctx, _NEW_PIXEL); - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { /* unpack pixelmap from PBO */ GLubyte *buf; /* Note, need to use DefaultPacking and Unpack's buffer object */ @@ -333,7 +333,7 @@ _mesa_PixelMapusv(GLenum map, GLsizei mapsize, const GLushort *values ) } } - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, ctx->Unpack.BufferObj); } @@ -359,7 +359,7 @@ _mesa_GetPixelMapfv( GLenum map, GLfloat *values ) mapsize = pm->Size; - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { /* pack pixelmap into PBO */ GLubyte *buf; /* Note, need to use DefaultPacking and Pack's buffer object */ @@ -397,7 +397,7 @@ _mesa_GetPixelMapfv( GLenum map, GLfloat *values ) MEMCPY(values, pm->Map, mapsize * sizeof(GLfloat)); } - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_PACK_BUFFER_EXT, ctx->Pack.BufferObj); } @@ -420,7 +420,7 @@ _mesa_GetPixelMapuiv( GLenum map, GLuint *values ) } mapsize = pm->Size; - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { /* pack pixelmap into PBO */ GLubyte *buf; /* Note, need to use DefaultPacking and Pack's buffer object */ @@ -458,7 +458,7 @@ _mesa_GetPixelMapuiv( GLenum map, GLuint *values ) } } - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_PACK_BUFFER_EXT, ctx->Pack.BufferObj); } @@ -481,7 +481,7 @@ _mesa_GetPixelMapusv( GLenum map, GLushort *values ) } mapsize = pm ? pm->Size : 0; - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { /* pack pixelmap into PBO */ GLubyte *buf; /* Note, need to use DefaultPacking and Pack's buffer object */ @@ -528,7 +528,7 @@ _mesa_GetPixelMapusv( GLenum map, GLushort *values ) } } - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_PACK_BUFFER_EXT, ctx->Pack.BufferObj); } diff --git a/src/mesa/main/polygon.c b/src/mesa/main/polygon.c index 564250b881c..d11c9424d5e 100644 --- a/src/mesa/main/polygon.c +++ b/src/mesa/main/polygon.c @@ -193,7 +193,7 @@ _mesa_PolygonMode( GLenum face, GLenum mode ) void _mesa_polygon_stipple(GLcontext *ctx, const GLubyte *pattern) { - if (ctx->Unpack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Unpack.BufferObj)) { /* Get/unpack the stipple pattern from a PBO */ GLubyte *buf; if (!_mesa_validate_pbo_access(2, &ctx->Unpack, 32, 32, 1, @@ -258,7 +258,7 @@ _mesa_GetPolygonStipple( GLubyte *dest ) /* XXX someday we may put this code into a separate function and call * it with ctx->Driver.GetPolygonStipple(). */ - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { /* Put/pack the stipple pattern into a PBO */ GLubyte *buf; if (!_mesa_validate_pbo_access(2, &ctx->Pack, 32, 32, 1, diff --git a/src/mesa/main/readpix.c b/src/mesa/main/readpix.c index 2326776ecbf..feea1d375f6 100644 --- a/src/mesa/main/readpix.c +++ b/src/mesa/main/readpix.c @@ -44,6 +44,10 @@ _mesa_error_check_format_type(GLcontext *ctx, GLenum format, GLenum type, GLboolean drawing) { const char *readDraw = drawing ? "Draw" : "Read"; + const GLboolean reading = !drawing; + + /* state validation should have already been done */ + ASSERT(ctx->NewState == 0x0); if (ctx->Extensions.EXT_packed_depth_stencil && type == GL_UNSIGNED_INT_24_8_EXT @@ -73,32 +77,45 @@ _mesa_error_check_format_type(GLcontext *ctx, GLenum format, GLenum type, case GL_RGBA: case GL_BGRA: case GL_ABGR_EXT: - if (drawing && !ctx->Visual.rgbMode) { - _mesa_error(ctx, GL_INVALID_OPERATION, + if (drawing) { + if (!ctx->DrawBuffer->Visual.rgbMode) { + _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawPixels(drawing RGB pixels into color index buffer)"); - return GL_TRUE; + return GL_TRUE; + } } - if (!drawing && !_mesa_dest_buffer_exists(ctx, GL_COLOR)) { - _mesa_error(ctx, GL_INVALID_OPERATION, - "glReadPixels(no color buffer)"); - return GL_TRUE; + else { + /* reading */ + if (!_mesa_source_buffer_exists(ctx, GL_COLOR)) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "glReadPixels(no color buffer)"); + return GL_TRUE; + } } break; case GL_COLOR_INDEX: - if (!drawing && ctx->Visual.rgbMode) { - _mesa_error(ctx, GL_INVALID_OPERATION, - "glReadPixels(reading color index format from RGB buffer)"); - return GL_TRUE; + if (drawing) { + if (ctx->DrawBuffer->Visual.rgbMode && + (ctx->PixelMaps.ItoR.Size == 0 || + ctx->PixelMaps.ItoG.Size == 0 || + ctx->PixelMaps.ItoB.Size == 0)) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "glDrawPixels(drawing color index pixels into RGB buffer)"); + return GL_TRUE; + } } - if (!drawing && !_mesa_dest_buffer_exists(ctx, GL_COLOR)) { - _mesa_error(ctx, GL_INVALID_OPERATION, - "glReadPixels(no color buffer)"); - return GL_TRUE; + else { + /* reading */ + if (!_mesa_source_buffer_exists(ctx, GL_COLOR)) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "glReadPixels(no color buffer)"); + return GL_TRUE; + } } break; case GL_STENCIL_INDEX: if ((drawing && !_mesa_dest_buffer_exists(ctx, format)) || - (!drawing && !_mesa_source_buffer_exists(ctx, format))) { + (reading && !_mesa_source_buffer_exists(ctx, format))) { _mesa_error(ctx, GL_INVALID_OPERATION, "gl%sPixels(no stencil buffer)", readDraw); return GL_TRUE; @@ -118,7 +135,7 @@ _mesa_error_check_format_type(GLcontext *ctx, GLenum format, GLenum type, return GL_TRUE; } if ((drawing && !_mesa_dest_buffer_exists(ctx, format)) || - (!drawing && !_mesa_source_buffer_exists(ctx, format))) { + (reading && !_mesa_source_buffer_exists(ctx, format))) { _mesa_error(ctx, GL_INVALID_OPERATION, "gl%sPixels(no depth or stencil buffer)", readDraw); return GL_TRUE; @@ -173,7 +190,7 @@ _mesa_ReadPixels( GLint x, GLint y, GLsizei width, GLsizei height, if (width == 0 || height == 0) return; /* nothing to do */ - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { if (!_mesa_validate_pbo_access(2, &ctx->Pack, width, height, 1, format, type, pixels)) { _mesa_error(ctx, GL_INVALID_OPERATION, @@ -181,7 +198,7 @@ _mesa_ReadPixels( GLint x, GLint y, GLsizei width, GLsizei height, return; } - if (ctx->Pack.BufferObj->Pointer) { + if (_mesa_bufferobj_mapped(ctx->Pack.BufferObj)) { /* buffer is mapped - that's an error */ _mesa_error(ctx, GL_INVALID_OPERATION, "glReadPixels(PBO is mapped)"); return; diff --git a/src/mesa/main/shared.c b/src/mesa/main/shared.c index ad6e6ce7cd3..93bbccd3c7f 100644 --- a/src/mesa/main/shared.c +++ b/src/mesa/main/shared.c @@ -196,7 +196,7 @@ delete_bufferobj_cb(GLuint id, void *data, void *userData) { struct gl_buffer_object *bufObj = (struct gl_buffer_object *) data; GLcontext *ctx = (GLcontext *) userData; - if (bufObj->Pointer) { + if (_mesa_bufferobj_mapped(bufObj)) { ctx->Driver.UnmapBuffer(ctx, 0, bufObj); bufObj->Pointer = NULL; } diff --git a/src/mesa/main/state.c b/src/mesa/main/state.c index d8191ab5187..140a998df2e 100644 --- a/src/mesa/main/state.c +++ b/src/mesa/main/state.c @@ -547,7 +547,8 @@ _mesa_update_state_locked( GLcontext *ctx ) /* Determine which state flags effect vertex/fragment program state */ if (ctx->FragmentProgram._MaintainTexEnvProgram) { prog_flags |= (_NEW_TEXTURE | _NEW_FOG | - _NEW_ARRAY | _NEW_LIGHT | _NEW_POINT | _NEW_RENDERMODE); + _NEW_ARRAY | _NEW_LIGHT | _NEW_POINT | _NEW_RENDERMODE | + _NEW_PROGRAM); } if (ctx->VertexProgram._MaintainTnlProgram) { prog_flags |= (_NEW_ARRAY | _NEW_TEXTURE | _NEW_TEXTURE_MATRIX | @@ -712,6 +713,6 @@ _mesa_set_vp_override(GLcontext *ctx, GLboolean flag) /* Set one of the bits which will trigger fragment program * regeneration: */ - ctx->NewState |= _NEW_ARRAY; + ctx->NewState |= _NEW_PROGRAM; } } diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c index 6b090ff399b..3736138b9ed 100644 --- a/src/mesa/main/texenvprogram.c +++ b/src/mesa/main/texenvprogram.c @@ -275,6 +275,7 @@ static GLbitfield get_fp_input_mask( GLcontext *ctx ) { /* _NEW_PROGRAM */ const GLboolean vertexShader = (ctx->Shader.CurrentProgram && + ctx->Shader.CurrentProgram->LinkStatus && ctx->Shader.CurrentProgram->VertexProgram); const GLboolean vertexProgram = ctx->VertexProgram._Enabled; GLbitfield fp_inputs = 0x0; @@ -321,8 +322,10 @@ static GLbitfield get_fp_input_mask( GLcontext *ctx ) /* Then look at what might be varying as a result of enabled * arrays, etc: */ - if (varying_inputs & VERT_BIT_COLOR0) fp_inputs |= FRAG_BIT_COL0; - if (varying_inputs & VERT_BIT_COLOR1) fp_inputs |= FRAG_BIT_COL1; + if (varying_inputs & VERT_BIT_COLOR0) + fp_inputs |= FRAG_BIT_COL0; + if (varying_inputs & VERT_BIT_COLOR1) + fp_inputs |= FRAG_BIT_COL1; fp_inputs |= (((varying_inputs & VERT_BIT_TEX_ANY) >> VERT_ATTRIB_TEX0) << FRAG_ATTRIB_TEX0); @@ -340,7 +343,7 @@ static GLbitfield get_fp_input_mask( GLcontext *ctx ) if (vertexShader) vprog = ctx->Shader.CurrentProgram->VertexProgram; else - vprog = ctx->VertexProgram._Current; + vprog = ctx->VertexProgram.Current; vp_outputs = vprog->Base.OutputsWritten; @@ -351,8 +354,10 @@ static GLbitfield get_fp_input_mask( GLcontext *ctx ) if (ctx->Point.PointSprite) vp_outputs |= FRAG_BITS_TEX_ANY; - if (vp_outputs & (1 << VERT_RESULT_COL0)) fp_inputs |= FRAG_BIT_COL0; - if (vp_outputs & (1 << VERT_RESULT_COL1)) fp_inputs |= FRAG_BIT_COL1; + if (vp_outputs & (1 << VERT_RESULT_COL0)) + fp_inputs |= FRAG_BIT_COL0; + if (vp_outputs & (1 << VERT_RESULT_COL1)) + fp_inputs |= FRAG_BIT_COL1; fp_inputs |= (((vp_outputs & VERT_RESULT_TEX_ANY) >> VERT_RESULT_TEX0) << FRAG_ATTRIB_TEX0); diff --git a/src/mesa/main/texgetimage.c b/src/mesa/main/texgetimage.c index 02409d80098..5557b694e37 100644 --- a/src/mesa/main/texgetimage.c +++ b/src/mesa/main/texgetimage.c @@ -30,6 +30,7 @@ #include "glheader.h" +#include "bufferobj.h" #include "context.h" #include "image.h" #include "texcompress.h" @@ -116,7 +117,7 @@ _mesa_get_teximage(GLcontext *ctx, GLenum target, GLint level, { const GLuint dimensions = (target == GL_TEXTURE_3D) ? 3 : 2; - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { /* Packing texture image into a PBO. * Map the (potentially) VRAM-based buffer into our process space so * we can write into it with the code below. @@ -296,7 +297,7 @@ _mesa_get_teximage(GLcontext *ctx, GLenum target, GLint level, } /* img */ } - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_PACK_BUFFER_EXT, ctx->Pack.BufferObj); } @@ -316,7 +317,7 @@ _mesa_get_compressed_teximage(GLcontext *ctx, GLenum target, GLint level, { GLuint size; - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { /* pack texture image into a PBO */ GLubyte *buf; if ((const GLubyte *) img + texImage->CompressedSize > @@ -349,7 +350,7 @@ _mesa_get_compressed_teximage(GLcontext *ctx, GLenum target, GLint level, /* just memcpy, no pixelstore or pixel transfer */ _mesa_memcpy(img, texImage->Data, size); - if (ctx->Pack.BufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_PACK_BUFFER_EXT, ctx->Pack.BufferObj); } diff --git a/src/mesa/main/teximage.c b/src/mesa/main/teximage.c index 6e21066537b..c758462a46a 100644 --- a/src/mesa/main/teximage.c +++ b/src/mesa/main/teximage.c @@ -1,6 +1,6 @@ /* * mesa 3-D graphics library - * Version: 7.5 + * Version: 7.6 * * Copyright (C) 1999-2008 Brian Paul All Rights Reserved. * Copyright (C) 2009 VMware, Inc. All Rights Reserved. @@ -679,37 +679,14 @@ _mesa_set_tex_image(struct gl_texture_object *tObj, GLenum target, GLint level, struct gl_texture_image *texImage) { + const GLuint face = _mesa_tex_target_to_face(target); + ASSERT(tObj); ASSERT(texImage); - /* XXX simplify this with _mesa_tex_target_to_face() */ - switch (target) { - case GL_TEXTURE_1D: - case GL_TEXTURE_2D: - case GL_TEXTURE_3D: - case GL_TEXTURE_1D_ARRAY_EXT: - case GL_TEXTURE_2D_ARRAY_EXT: - tObj->Image[0][level] = texImage; - break; - case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB: - case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB: - case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB: - { - GLuint face = ((GLuint) target - - (GLuint) GL_TEXTURE_CUBE_MAP_POSITIVE_X); - tObj->Image[face][level] = texImage; - } - break; - case GL_TEXTURE_RECTANGLE_NV: - ASSERT(level == 0); - tObj->Image[0][level] = texImage; - break; - default: - _mesa_problem(NULL, "bad target in _mesa_set_tex_image()"); - return; - } + ASSERT(target != GL_TEXTURE_RECTANGLE_NV || level == 0); + + tObj->Image[face][level] = texImage; + /* Set the 'back' pointer */ texImage->TexObject = tObj; } @@ -788,6 +765,9 @@ _mesa_delete_texture_image( GLcontext *ctx, struct gl_texture_image *texImage ) GLboolean _mesa_is_proxy_texture(GLenum target) { + /* NUM_TEXTURE_TARGETS should match number of terms below */ + assert(NUM_TEXTURE_TARGETS == 7); + return (target == GL_PROXY_TEXTURE_1D || target == GL_PROXY_TEXTURE_2D || target == GL_PROXY_TEXTURE_3D || @@ -799,6 +779,18 @@ _mesa_is_proxy_texture(GLenum target) /** + * Return pointer to current texture unit. + * This the texture unit set by glActiveTexture(), not glClientActiveTexture(). + */ +static INLINE struct gl_texture_unit * +get_current_tex_unit(GLcontext *ctx) +{ + ASSERT(ctx->Texture.CurrentUnit < Elements(ctx->Texture.Unit)); + return &(ctx->Texture.Unit[ctx->Texture.CurrentUnit]); +} + + +/** * Get the texture object that corresponds to the target of the given texture unit. * * \param ctx GL context. @@ -864,74 +856,28 @@ _mesa_select_tex_object(GLcontext *ctx, const struct gl_texture_unit *texUnit, /** - * Get the texture image struct which corresponds to target and level - * of the given texture unit. + * Get a texture image pointer from a texture object, given a texture + * target and mipmap level. The target and level parameters should + * have already been error-checked. * * \param ctx GL context. * \param texObj texture unit. * \param target texture target. * \param level image level. * - * \return pointer to the texture image structure on success, or NULL on failure. - * - * \sa gl_texture_unit. + * \return pointer to the texture image structure, or NULL on failure. */ struct gl_texture_image * _mesa_select_tex_image(GLcontext *ctx, const struct gl_texture_object *texObj, GLenum target, GLint level) { - ASSERT(texObj); - - if (level < 0 || level >= MAX_TEXTURE_LEVELS) - return NULL; - - /* XXX simplify this with _mesa_tex_target_to_face() */ - switch (target) { - case GL_TEXTURE_1D: - case GL_PROXY_TEXTURE_1D: - case GL_TEXTURE_2D: - case GL_PROXY_TEXTURE_2D: - case GL_TEXTURE_3D: - case GL_PROXY_TEXTURE_3D: - return texObj->Image[0][level]; - - case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB: - case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB: - case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB: - case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB: - if (ctx->Extensions.ARB_texture_cube_map) { - GLuint face = ((GLuint) target - - (GLuint) GL_TEXTURE_CUBE_MAP_POSITIVE_X); - return texObj->Image[face][level]; - } - else - return NULL; - - case GL_PROXY_TEXTURE_CUBE_MAP_ARB: - if (ctx->Extensions.ARB_texture_cube_map) - return texObj->Image[0][level]; - else - return NULL; - - case GL_TEXTURE_RECTANGLE_NV: - case GL_PROXY_TEXTURE_RECTANGLE_NV: - if (ctx->Extensions.NV_texture_rectangle && level == 0) - return texObj->Image[0][level]; - else - return NULL; + const GLuint face = _mesa_tex_target_to_face(target); - case GL_TEXTURE_1D_ARRAY_EXT: - case GL_PROXY_TEXTURE_1D_ARRAY_EXT: - case GL_TEXTURE_2D_ARRAY_EXT: - case GL_PROXY_TEXTURE_2D_ARRAY_EXT: - return (ctx->Extensions.MESA_texture_array) - ? texObj->Image[0][level] : NULL; + ASSERT(texObj); + ASSERT(level >= 0); + ASSERT(level < MAX_TEXTURE_LEVELS); - default: - return NULL; - } + return texObj->Image[face][level]; } @@ -1273,6 +1219,23 @@ _mesa_init_teximage_fields(GLcontext *ctx, GLenum target, /** + * Free and clear fields of the gl_texture_image struct. + * + * \param ctx GL context. + * \param texImage texture image structure to be cleared. + * + * After the call, \p texImage will have no data associated with it. Its + * fields are cleared so that its parent object will test incomplete. + */ +void +_mesa_clear_texture_image(GLcontext *ctx, struct gl_texture_image *texImage) +{ + ctx->Driver.FreeTexImageData(ctx, texImage); + clear_teximage_fields(texImage); +} + + +/** * This is the fallback for Driver.TestProxyTexImage(). Test the texture * level, width, height and depth against the ctx->Const limits for textures. * @@ -1926,14 +1889,14 @@ copytexture_error_check( GLcontext *ctx, GLuint dimensions, return GL_TRUE; } - /* NOTE: the format and type aren't really significant for - * TestProxyTexImage(). Only the internalformat really matters. if (!_mesa_source_buffer_exists(ctx, format)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glCopyTexImage%dD(missing readbuffer)", dimensions); return GL_TRUE; } + /* NOTE: the format and type aren't really significant for + * TestProxyTexImage(). Only the internalformat really matters. */ type = GL_FLOAT; @@ -2261,142 +2224,165 @@ copytexsubimage_error_check2( GLcontext *ctx, GLuint dimensions, /** - * Get texture image. Called by glGetTexImage. - * - * \param target texture target. - * \param level image level. - * \param format pixel data format for returned image. - * \param type pixel data type for returned image. - * \param pixels returned pixel data. + * Do error checking for a glGetTexImage() call. + * \return GL_TRUE if any error, GL_FALSE if no errors. */ -void GLAPIENTRY -_mesa_GetTexImage( GLenum target, GLint level, GLenum format, - GLenum type, GLvoid *pixels ) +static GLboolean +getteximage_error_check(GLcontext *ctx, GLenum target, GLint level, + GLenum format, GLenum type, GLvoid *pixels ) { const struct gl_texture_unit *texUnit; struct gl_texture_object *texObj; struct gl_texture_image *texImage; - GLint maxLevels = 0; - GET_CURRENT_CONTEXT(ctx); - ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); - - texUnit = &(ctx->Texture.Unit[ctx->Texture.CurrentUnit]); - texObj = _mesa_select_tex_object(ctx, texUnit, target); - if (!texObj || _mesa_is_proxy_texture(target)) { - _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexImage(target)"); - return; - } + const GLuint maxLevels = _mesa_max_texture_levels(ctx, target); - maxLevels = _mesa_max_texture_levels(ctx, target); ASSERT(maxLevels > 0); /* 0 indicates bad target, caught above */ if (level < 0 || level >= maxLevels) { _mesa_error( ctx, GL_INVALID_VALUE, "glGetTexImage(level)" ); - return; + return GL_TRUE; } if (_mesa_sizeof_packed_type(type) <= 0) { _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexImage(type)" ); - return; + return GL_TRUE; } if (_mesa_components_in_format(format) <= 0 || format == GL_STENCIL_INDEX) { _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexImage(format)" ); - return; + return GL_TRUE; } if (!ctx->Extensions.EXT_paletted_texture && is_index_format(format)) { _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexImage(format)"); - return; + return GL_TRUE; } if (!ctx->Extensions.ARB_depth_texture && is_depth_format(format)) { _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexImage(format)"); - return; + return GL_TRUE; } if (!ctx->Extensions.MESA_ycbcr_texture && is_ycbcr_format(format)) { _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexImage(format)"); - return; + return GL_TRUE; } if (!ctx->Extensions.EXT_packed_depth_stencil && is_depthstencil_format(format)) { _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexImage(format)"); - return; + return GL_TRUE; } if (!ctx->Extensions.ATI_envmap_bumpmap && is_dudv_format(format)) { _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexImage(format)"); - return; + return GL_TRUE; } - _mesa_lock_texture(ctx, texObj); - { - texImage = _mesa_select_tex_image(ctx, texObj, target, level); - if (!texImage) { - /* invalid mipmap level, not an error */ - goto out; - } + texUnit = get_current_tex_unit(ctx); + texObj = _mesa_select_tex_object(ctx, texUnit, target); + if (!texObj || _mesa_is_proxy_texture(target)) { + _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexImage(target)"); + return GL_TRUE; + } - /* Make sure the requested image format is compatible with the - * texture's format. Note that a color index texture can be converted - * to RGBA so that combo is allowed. - */ - if (_mesa_is_color_format(format) - && !_mesa_is_color_format(texImage->TexFormat->BaseFormat) - && !is_index_format(texImage->TexFormat->BaseFormat)) { - _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)"); - goto out; - } - else if (is_index_format(format) - && !is_index_format(texImage->TexFormat->BaseFormat)) { - _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)"); - goto out; - } - else if (is_depth_format(format) - && !is_depth_format(texImage->TexFormat->BaseFormat) - && !is_depthstencil_format(texImage->TexFormat->BaseFormat)) { - _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)"); - goto out; - } - else if (is_ycbcr_format(format) - && !is_ycbcr_format(texImage->TexFormat->BaseFormat)) { - _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)"); - goto out; - } - else if (is_depthstencil_format(format) - && !is_depthstencil_format(texImage->TexFormat->BaseFormat)) { - _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)"); - goto out; - } - else if (is_dudv_format(format) - && !is_dudv_format(texImage->TexFormat->BaseFormat)) { - _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)"); - goto out; - } + texImage = _mesa_select_tex_image(ctx, texObj, target, level); + if (!texImage) { + /* out of memory */ + return GL_TRUE; + } + + /* Make sure the requested image format is compatible with the + * texture's format. Note that a color index texture can be converted + * to RGBA so that combo is allowed. + */ + if (_mesa_is_color_format(format) + && !_mesa_is_color_format(texImage->TexFormat->BaseFormat) + && !is_index_format(texImage->TexFormat->BaseFormat)) { + _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)"); + return GL_TRUE; + } + else if (is_index_format(format) + && !is_index_format(texImage->TexFormat->BaseFormat)) { + _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)"); + return GL_TRUE; + } + else if (is_depth_format(format) + && !is_depth_format(texImage->TexFormat->BaseFormat) + && !is_depthstencil_format(texImage->TexFormat->BaseFormat)) { + _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)"); + return GL_TRUE; + } + else if (is_ycbcr_format(format) + && !is_ycbcr_format(texImage->TexFormat->BaseFormat)) { + _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)"); + return GL_TRUE; + } + else if (is_depthstencil_format(format) + && !is_depthstencil_format(texImage->TexFormat->BaseFormat)) { + _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)"); + return GL_TRUE; + } + else if (is_dudv_format(format) + && !is_dudv_format(texImage->TexFormat->BaseFormat)) { + _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)"); + return GL_TRUE; + } - if (ctx->Pack.BufferObj->Name) { - /* packing texture image into a PBO */ - const GLuint dimensions = (target == GL_TEXTURE_3D) ? 3 : 2; - if (!_mesa_validate_pbo_access(dimensions, &ctx->Pack, texImage->Width, - texImage->Height, texImage->Depth, - format, type, pixels)) { - _mesa_error(ctx, GL_INVALID_OPERATION, - "glGetTexImage(invalid PBO access)"); - goto out; - } + if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) { + /* packing texture image into a PBO */ + const GLuint dimensions = (target == GL_TEXTURE_3D) ? 3 : 2; + if (!_mesa_validate_pbo_access(dimensions, &ctx->Pack, texImage->Width, + texImage->Height, texImage->Depth, + format, type, pixels)) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "glGetTexImage(invalid PBO access)"); + return GL_TRUE; } + } + + return GL_FALSE; +} + + + +/** + * Get texture image. Called by glGetTexImage. + * + * \param target texture target. + * \param level image level. + * \param format pixel data format for returned image. + * \param type pixel data type for returned image. + * \param pixels returned pixel data. + */ +void GLAPIENTRY +_mesa_GetTexImage( GLenum target, GLint level, GLenum format, + GLenum type, GLvoid *pixels ) +{ + const struct gl_texture_unit *texUnit; + struct gl_texture_object *texObj; + GET_CURRENT_CONTEXT(ctx); + ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + + if (getteximage_error_check(ctx, target, level, format, type, pixels)) { + return; + } + + texUnit = get_current_tex_unit(ctx); + texObj = _mesa_select_tex_object(ctx, texUnit, target); + + _mesa_lock_texture(ctx, texObj); + { + struct gl_texture_image *texImage = + _mesa_select_tex_image(ctx, texObj, target, level); /* typically, this will call _mesa_get_teximage() */ ctx->Driver.GetTexImage(ctx, target, level, format, type, pixels, - texObj, texImage); - + texObj, texImage); } - out: _mesa_unlock_texture(ctx, texObj); } @@ -2544,7 +2530,7 @@ _mesa_TexImage1D( GLenum target, GLint level, GLint internalFormat, if (ctx->NewState & _MESA_NEW_TRANSFER_STATE) _mesa_update_state(ctx); - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); { @@ -2652,7 +2638,7 @@ _mesa_TexImage2D( GLenum target, GLint level, GLint internalFormat, if (ctx->NewState & _MESA_NEW_TRANSFER_STATE) _mesa_update_state(ctx); - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); { @@ -2755,7 +2741,7 @@ _mesa_TexImage3D( GLenum target, GLint level, GLint internalFormat, if (ctx->NewState & _MESA_NEW_TRANSFER_STATE) _mesa_update_state(ctx); - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); { @@ -2861,7 +2847,7 @@ _mesa_TexSubImage1D( GLenum target, GLint level, } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); assert(texObj); @@ -2921,7 +2907,7 @@ _mesa_TexSubImage2D( GLenum target, GLint level, return; /* error was detected */ } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); { @@ -2973,7 +2959,7 @@ _mesa_TexSubImage3D( GLenum target, GLint level, return; /* error was detected */ } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); @@ -3034,7 +3020,7 @@ _mesa_CopyTexImage1D( GLenum target, GLint level, postConvWidth, 1, border)) return; - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); { @@ -3100,7 +3086,7 @@ _mesa_CopyTexImage2D( GLenum target, GLint level, GLenum internalFormat, postConvWidth, postConvHeight, border)) return; - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); @@ -3160,7 +3146,7 @@ _mesa_CopyTexSubImage1D( GLenum target, GLint level, if (copytexsubimage_error_check1(ctx, 1, target, level)) return; - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); @@ -3215,7 +3201,7 @@ _mesa_CopyTexSubImage2D( GLenum target, GLint level, if (copytexsubimage_error_check1(ctx, 2, target, level)) return; - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); @@ -3270,7 +3256,7 @@ _mesa_CopyTexSubImage3D( GLenum target, GLint level, if (copytexsubimage_error_check1(ctx, 3, target, level)) return; - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); @@ -3524,7 +3510,7 @@ _mesa_CompressedTexImage1DARB(GLenum target, GLint level, return; } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); @@ -3578,7 +3564,7 @@ _mesa_CompressedTexImage1DARB(GLenum target, GLint level, struct gl_texture_unit *texUnit; struct gl_texture_object *texObj; struct gl_texture_image *texImage; - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); @@ -3621,7 +3607,7 @@ _mesa_CompressedTexImage2DARB(GLenum target, GLint level, return; } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); @@ -3677,7 +3663,7 @@ _mesa_CompressedTexImage2DARB(GLenum target, GLint level, struct gl_texture_unit *texUnit; struct gl_texture_object *texObj; struct gl_texture_image *texImage; - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); @@ -3717,7 +3703,7 @@ _mesa_CompressedTexImage3DARB(GLenum target, GLint level, return; } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); { @@ -3771,7 +3757,7 @@ _mesa_CompressedTexImage3DARB(GLenum target, GLint level, struct gl_texture_unit *texUnit; struct gl_texture_object *texObj; struct gl_texture_image *texImage; - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); { @@ -3810,7 +3796,7 @@ _mesa_CompressedTexSubImage1DARB(GLenum target, GLint level, GLint xoffset, return; } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); { @@ -3867,7 +3853,7 @@ _mesa_CompressedTexSubImage2DARB(GLenum target, GLint level, GLint xoffset, return; } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); { @@ -3924,7 +3910,7 @@ _mesa_CompressedTexSubImage3DARB(GLenum target, GLint level, GLint xoffset, return; } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); { @@ -3971,7 +3957,7 @@ _mesa_GetCompressedTexImageARB(GLenum target, GLint level, GLvoid *img) GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); if (!texObj) { _mesa_error(ctx, GL_INVALID_ENUM, "glGetCompressedTexImageARB"); diff --git a/src/mesa/main/teximage.h b/src/mesa/main/teximage.h index eb60a1fa8fa..b0d7c1c3aa1 100644 --- a/src/mesa/main/teximage.h +++ b/src/mesa/main/teximage.h @@ -73,6 +73,10 @@ _mesa_init_teximage_fields(GLcontext *ctx, GLenum target, extern void +_mesa_clear_texture_image(GLcontext *ctx, struct gl_texture_image *texImage); + + +extern void _mesa_set_tex_image(struct gl_texture_object *tObj, GLenum target, GLint level, struct gl_texture_image *texImage); diff --git a/src/mesa/main/texobj.c b/src/mesa/main/texobj.c index 2082f945f18..22657ed81b7 100644 --- a/src/mesa/main/texobj.c +++ b/src/mesa/main/texobj.c @@ -261,6 +261,32 @@ _mesa_copy_texture_object( struct gl_texture_object *dest, /** + * Clear all texture images of the given texture object. + * + * \param ctx GL context. + * \param t texture object. + * + * \sa _mesa_clear_texture_image(). + */ +void +_mesa_clear_texture_object(GLcontext *ctx, struct gl_texture_object *texObj) +{ + GLuint i, j; + + if (texObj->Target == 0) + return; + + for (i = 0; i < MAX_FACES; i++) { + for (j = 0; j < MAX_TEXTURE_LEVELS; j++) { + struct gl_texture_image *texImage = texObj->Image[i][j]; + if (texImage) + _mesa_clear_texture_image(ctx, texImage); + } + } +} + + +/** * Check if the given texture object is valid by examining its Target field. * For debugging only. */ @@ -665,6 +691,24 @@ _mesa_test_texobj_completeness( const GLcontext *ctx, /** + * Mark a texture object dirty. It forces the object to be incomplete + * and optionally forces the context to re-validate its state. + * + * \param ctx GL context. + * \param texObj texture object. + * \param invalidate_state also invalidate context state. + */ +void +_mesa_dirty_texobj(GLcontext *ctx, struct gl_texture_object *texObj, + GLboolean invalidate_state) +{ + texObj->_Complete = GL_FALSE; + if (invalidate_state) + ctx->NewState |= _NEW_TEXTURE; +} + + +/** * Return pointer to a default/fallback texture. * The texture is a 2D 8x8 RGBA texture with all texels = (0,0,0,1). * That's the value a sampler should get when sampling from an @@ -715,7 +759,6 @@ _mesa_get_fallback_texture(GLcontext *ctx) } - /*@}*/ @@ -1160,10 +1203,9 @@ _mesa_IsTexture( GLuint texture ) /** - * Simplest implementation of texture locking: Grab the a new mutex in - * the shared context. Examine the shared context state timestamp and - * if there has been a change, set the appropriate bits in - * ctx->NewState. + * Simplest implementation of texture locking: grab the shared tex + * mutex. Examine the shared context state timestamp and if there has + * been a change, set the appropriate bits in ctx->NewState. * * This is used to deal with synchronizing things when a texture object * is used/modified by different contexts (or threads) which are sharing diff --git a/src/mesa/main/texobj.h b/src/mesa/main/texobj.h index 2599c0816a9..9bfebd45c81 100644 --- a/src/mesa/main/texobj.h +++ b/src/mesa/main/texobj.h @@ -58,6 +58,9 @@ _mesa_copy_texture_object( struct gl_texture_object *dest, const struct gl_texture_object *src ); extern void +_mesa_clear_texture_object(GLcontext *ctx, struct gl_texture_object *obj); + +extern void _mesa_reference_texobj(struct gl_texture_object **ptr, struct gl_texture_object *tex); @@ -65,6 +68,10 @@ extern void _mesa_test_texobj_completeness( const GLcontext *ctx, struct gl_texture_object *obj ); +extern void +_mesa_dirty_texobj(GLcontext *ctx, struct gl_texture_object *texObj, + GLboolean invalidate_state); + extern struct gl_texture_object * _mesa_get_fallback_texture(GLcontext *ctx); @@ -76,7 +83,6 @@ _mesa_lock_context_textures( GLcontext *ctx ); /*@}*/ - /** * \name API functions */ diff --git a/src/mesa/main/texstore.c b/src/mesa/main/texstore.c index bfced1b3f4f..a22db628d3e 100644 --- a/src/mesa/main/texstore.c +++ b/src/mesa/main/texstore.c @@ -3138,7 +3138,7 @@ _mesa_validate_pbo_teximage(GLcontext *ctx, GLuint dimensions, { GLubyte *buf; - if (unpack->BufferObj->Name == 0) { + if (!_mesa_is_bufferobj(unpack->BufferObj)) { /* no PBO */ return pixels; } @@ -3174,7 +3174,7 @@ _mesa_validate_pbo_compressed_teximage(GLcontext *ctx, { GLubyte *buf; - if (packing->BufferObj->Name == 0) { + if (!_mesa_is_bufferobj(packing->BufferObj)) { /* not using a PBO - return pointer unchanged */ return pixels; } @@ -3204,7 +3204,7 @@ void _mesa_unmap_teximage_pbo(GLcontext *ctx, const struct gl_pixelstore_attrib *unpack) { - if (unpack->BufferObj->Name) { + if (_mesa_is_bufferobj(unpack->BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT, unpack->BufferObj); } diff --git a/src/mesa/main/varray.c b/src/mesa/main/varray.c index 3d5b8faecff..be1c03cec2a 100644 --- a/src/mesa/main/varray.c +++ b/src/mesa/main/varray.c @@ -135,7 +135,8 @@ _mesa_VertexPointer(GLint size, GLenum type, GLsizei stride, const GLvoid *ptr) break; #endif default: - _mesa_error( ctx, GL_INVALID_ENUM, "glVertexPointer(type)" ); + _mesa_error( ctx, GL_INVALID_ENUM, "glVertexPointer(type=%s)", + _mesa_lookup_enum_by_nr(type)); return; } @@ -186,7 +187,8 @@ _mesa_NormalPointer(GLenum type, GLsizei stride, const GLvoid *ptr ) break; #endif default: - _mesa_error( ctx, GL_INVALID_ENUM, "glNormalPointer(type)" ); + _mesa_error( ctx, GL_INVALID_ENUM, "glNormalPointer(type=%s)", + _mesa_lookup_enum_by_nr(type)); return; } @@ -265,7 +267,8 @@ _mesa_ColorPointer(GLint size, GLenum type, GLsizei stride, const GLvoid *ptr) break; #endif default: - _mesa_error( ctx, GL_INVALID_ENUM, "glColorPointer(type)" ); + _mesa_error( ctx, GL_INVALID_ENUM, "glColorPointer(type=%s)", + _mesa_lookup_enum_by_nr(type)); return; } @@ -415,7 +418,8 @@ _mesa_SecondaryColorPointerEXT(GLint size, GLenum type, elementSize = size * sizeof(GLdouble); break; default: - _mesa_error( ctx, GL_INVALID_ENUM, "glSecondaryColorPointer(type)" ); + _mesa_error( ctx, GL_INVALID_ENUM, "glSecondaryColorPointer(type=%s)", + _mesa_lookup_enum_by_nr(type)); return; } @@ -476,7 +480,8 @@ _mesa_TexCoordPointer(GLint size, GLenum type, GLsizei stride, break; #endif default: - _mesa_error( ctx, GL_INVALID_ENUM, "glTexCoordPointer(type)" ); + _mesa_error( ctx, GL_INVALID_ENUM, "glTexCoordPointer(type=%s)", + _mesa_lookup_enum_by_nr(type)); return; } @@ -610,7 +615,8 @@ _mesa_VertexAttribPointerNV(GLuint index, GLint size, GLenum type, elementSize = size * sizeof(GLdouble); break; default: - _mesa_error( ctx, GL_INVALID_ENUM, "glVertexAttribPointerNV(type)" ); + _mesa_error( ctx, GL_INVALID_ENUM, "glVertexAttribPointerNV(type=%s)", + _mesa_lookup_enum_by_nr(type)); return; } @@ -1093,6 +1099,29 @@ _mesa_MultiModeDrawElementsIBM( const GLenum * mode, const GLsizei * count, /** + * Copy one client vertex array to another. + */ +void +_mesa_copy_client_array(GLcontext *ctx, + struct gl_client_array *dst, + struct gl_client_array *src) +{ + dst->Size = src->Size; + dst->Type = src->Type; + dst->Format = src->Format; + dst->Stride = src->Stride; + dst->StrideB = src->StrideB; + dst->Ptr = src->Ptr; + dst->Enabled = src->Enabled; + dst->Normalized = src->Normalized; + dst->_ElementSize = src->_ElementSize; + _mesa_reference_buffer_object(ctx, &dst->BufferObj, src->BufferObj); + dst->_MaxElement = src->_MaxElement; +} + + + +/** * Print vertex array's fields. */ static void diff --git a/src/mesa/main/varray.h b/src/mesa/main/varray.h index d4d505ae049..becc67c29d3 100644 --- a/src/mesa/main/varray.h +++ b/src/mesa/main/varray.h @@ -161,6 +161,12 @@ _mesa_DrawRangeElements(GLenum mode, GLuint start, GLuint end, GLsizei count, extern void +_mesa_copy_client_array(GLcontext *ctx, + struct gl_client_array *dst, + struct gl_client_array *src); + + +extern void _mesa_print_arrays(GLcontext *ctx); extern void diff --git a/src/mesa/shader/arbprogparse.c b/src/mesa/shader/arbprogparse.c index f428ee541b5..381445c0e84 100644 --- a/src/mesa/shader/arbprogparse.c +++ b/src/mesa/shader/arbprogparse.c @@ -2027,6 +2027,7 @@ parse_param (GLcontext * ctx, const GLubyte ** inst, struct var_cache **vc_head, if (specified_length != (int)param_var->param_binding_length) { program_error(ctx, Program->Position, "Declared parameter array length does not match parameter list"); + return 1; } } diff --git a/src/mesa/shader/prog_print.c b/src/mesa/shader/prog_print.c index cef0288a9c9..20d3fdec477 100644 --- a/src/mesa/shader/prog_print.c +++ b/src/mesa/shader/prog_print.c @@ -933,7 +933,7 @@ _mesa_write_shader_to_file(const struct gl_shader *shader) return; } - fprintf(f, "/* Shader %u source */\n", shader->Name); + fprintf(f, "/* Shader %u source, checksum %u */\n", shader->Name, shader->SourceChecksum); fputs(shader->Source, f); fprintf(f, "\n"); diff --git a/src/mesa/shader/shader_api.c b/src/mesa/shader/shader_api.c index 648fbc186bb..9cfc6dd81c5 100644 --- a/src/mesa/shader/shader_api.c +++ b/src/mesa/shader/shader_api.c @@ -1431,6 +1431,9 @@ _mesa_shader_source(GLcontext *ctx, GLuint shader, const GLchar *source) } sh->Source = source; sh->CompileStatus = GL_FALSE; +#ifdef DEBUG + sh->SourceChecksum = _mesa_str_checksum(sh->Source); +#endif } @@ -1506,9 +1509,10 @@ _mesa_use_program(GLcontext *ctx, GLuint program) GLuint i; _mesa_printf("Use Shader %u\n", shProg->Name); for (i = 0; i < shProg->NumShaders; i++) { - _mesa_printf(" shader %u, type 0x%x\n", + _mesa_printf(" shader %u, type 0x%x, checksum %u\n", shProg->Shaders[i]->Name, - shProg->Shaders[i]->Type); + shProg->Shaders[i]->Type, + shProg->Shaders[i]->SourceChecksum); } if (shProg->VertexProgram) printf(" vert prog %u\n", shProg->VertexProgram->Base.Id); diff --git a/src/mesa/shader/slang/slang_builtin.c b/src/mesa/shader/slang/slang_builtin.c index 289d94644f0..9ed8e3ecb39 100644 --- a/src/mesa/shader/slang/slang_builtin.c +++ b/src/mesa/shader/slang/slang_builtin.c @@ -710,3 +710,187 @@ _slang_alloc_statevar(slang_ir_node *n, *direct = GL_FALSE; return alloc_state_var_array(n->Var, paramList); } + + + + +#define SWIZZLE_ZWWW MAKE_SWIZZLE4(SWIZZLE_Z, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W) + + +/** Predefined shader inputs */ +struct input_info +{ + const char *Name; + GLuint Attrib; + GLenum Type; + GLuint Swizzle; +}; + +/** Predefined vertex shader inputs/attributes */ +static const struct input_info vertInputs[] = { + { "gl_Vertex", VERT_ATTRIB_POS, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { "gl_Normal", VERT_ATTRIB_NORMAL, GL_FLOAT_VEC3, SWIZZLE_NOOP }, + { "gl_Color", VERT_ATTRIB_COLOR0, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { "gl_SecondaryColor", VERT_ATTRIB_COLOR1, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { "gl_FogCoord", VERT_ATTRIB_FOG, GL_FLOAT, SWIZZLE_XXXX }, + { "gl_MultiTexCoord0", VERT_ATTRIB_TEX0, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { "gl_MultiTexCoord1", VERT_ATTRIB_TEX1, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { "gl_MultiTexCoord2", VERT_ATTRIB_TEX2, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { "gl_MultiTexCoord3", VERT_ATTRIB_TEX3, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { "gl_MultiTexCoord4", VERT_ATTRIB_TEX4, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { "gl_MultiTexCoord5", VERT_ATTRIB_TEX5, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { "gl_MultiTexCoord6", VERT_ATTRIB_TEX6, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { "gl_MultiTexCoord7", VERT_ATTRIB_TEX7, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { NULL, 0, SWIZZLE_NOOP } +}; + +/** Predefined fragment shader inputs */ +static const struct input_info fragInputs[] = { + { "gl_FragCoord", FRAG_ATTRIB_WPOS, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { "gl_Color", FRAG_ATTRIB_COL0, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { "gl_SecondaryColor", FRAG_ATTRIB_COL1, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + { "gl_TexCoord", FRAG_ATTRIB_TEX0, GL_FLOAT_VEC4, SWIZZLE_NOOP }, + /* note: we're packing several quantities into the fogcoord vector */ + { "gl_FogFragCoord", FRAG_ATTRIB_FOGC, GL_FLOAT, SWIZZLE_XXXX }, + { "gl_FogFragCoord", FRAG_ATTRIB_FOGC, GL_FLOAT, SWIZZLE_XXXX }, + { "gl_FrontFacing", FRAG_ATTRIB_FACE, GL_FLOAT, SWIZZLE_XXXX }, + { "gl_PointCoord", FRAG_ATTRIB_PNTC, GL_FLOAT_VEC2, SWIZZLE_XYZW }, + { NULL, 0, SWIZZLE_NOOP } +}; + + +/** + * Return the VERT_ATTRIB_* or FRAG_ATTRIB_* value that corresponds to + * a vertex or fragment program input variable. Return -1 if the input + * name is invalid. + * XXX return size too + */ +GLint +_slang_input_index(const char *name, GLenum target, GLuint *swizzleOut) +{ + const struct input_info *inputs; + GLuint i; + + switch (target) { + case GL_VERTEX_PROGRAM_ARB: + inputs = vertInputs; + break; + case GL_FRAGMENT_PROGRAM_ARB: + inputs = fragInputs; + break; + /* XXX geom program */ + default: + _mesa_problem(NULL, "bad target in _slang_input_index"); + return -1; + } + + ASSERT(MAX_TEXTURE_COORD_UNITS == 8); /* if this fails, fix vertInputs above */ + + for (i = 0; inputs[i].Name; i++) { + if (strcmp(inputs[i].Name, name) == 0) { + /* found */ + *swizzleOut = inputs[i].Swizzle; + return inputs[i].Attrib; + } + } + return -1; +} + + +/** + * Return name of the given vertex attribute (VERT_ATTRIB_x). + */ +const char * +_slang_vert_attrib_name(GLuint attrib) +{ + GLuint i; + assert(attrib < VERT_ATTRIB_GENERIC0); + for (i = 0; vertInputs[i].Name; i++) { + if (vertInputs[i].Attrib == attrib) + return vertInputs[i].Name; + } + return NULL; +} + + +/** + * Return type (GL_FLOAT, GL_FLOAT_VEC2, etc) of the given vertex + * attribute (VERT_ATTRIB_x). + */ +GLenum +_slang_vert_attrib_type(GLuint attrib) +{ + GLuint i; + assert(attrib < VERT_ATTRIB_GENERIC0); + for (i = 0; vertInputs[i].Name; i++) { + if (vertInputs[i].Attrib == attrib) + return vertInputs[i].Type; + } + return GL_NONE; +} + + + + + +/** Predefined shader output info */ +struct output_info +{ + const char *Name; + GLuint Attrib; +}; + +/** Predefined vertex shader outputs */ +static const struct output_info vertOutputs[] = { + { "gl_Position", VERT_RESULT_HPOS }, + { "gl_FrontColor", VERT_RESULT_COL0 }, + { "gl_BackColor", VERT_RESULT_BFC0 }, + { "gl_FrontSecondaryColor", VERT_RESULT_COL1 }, + { "gl_BackSecondaryColor", VERT_RESULT_BFC1 }, + { "gl_TexCoord", VERT_RESULT_TEX0 }, + { "gl_FogFragCoord", VERT_RESULT_FOGC }, + { "gl_PointSize", VERT_RESULT_PSIZ }, + { NULL, 0 } +}; + +/** Predefined fragment shader outputs */ +static const struct output_info fragOutputs[] = { + { "gl_FragColor", FRAG_RESULT_COLOR }, + { "gl_FragDepth", FRAG_RESULT_DEPTH }, + { "gl_FragData", FRAG_RESULT_DATA0 }, + { NULL, 0 } +}; + + +/** + * Return the VERT_RESULT_* or FRAG_RESULT_* value that corresponds to + * a vertex or fragment program output variable. Return -1 for an invalid + * output name. + */ +GLint +_slang_output_index(const char *name, GLenum target) +{ + const struct output_info *outputs; + GLuint i; + + switch (target) { + case GL_VERTEX_PROGRAM_ARB: + outputs = vertOutputs; + break; + case GL_FRAGMENT_PROGRAM_ARB: + outputs = fragOutputs; + break; + /* XXX geom program */ + default: + _mesa_problem(NULL, "bad target in _slang_output_index"); + return -1; + } + + for (i = 0; outputs[i].Name; i++) { + if (strcmp(outputs[i].Name, name) == 0) { + /* found */ + return outputs[i].Attrib; + } + } + return -1; +} diff --git a/src/mesa/shader/slang/slang_builtin.h b/src/mesa/shader/slang/slang_builtin.h index 7f6fe80fcca..f814d11ac73 100644 --- a/src/mesa/shader/slang/slang_builtin.h +++ b/src/mesa/shader/slang/slang_builtin.h @@ -37,4 +37,18 @@ _slang_alloc_statevar(slang_ir_node *n, GLboolean *direct); +extern GLint +_slang_input_index(const char *name, GLenum target, GLuint *swizzleOut); + +extern GLint +_slang_output_index(const char *name, GLenum target); + + +extern const char * +_slang_vert_attrib_name(GLuint attrib); + +extern GLenum +_slang_vert_attrib_type(GLuint attrib); + + #endif /* SLANG_BUILTIN_H */ diff --git a/src/mesa/shader/slang/slang_codegen.c b/src/mesa/shader/slang/slang_codegen.c index 4fe68eafa7f..349f432deca 100644 --- a/src/mesa/shader/slang/slang_codegen.c +++ b/src/mesa/shader/slang/slang_codegen.c @@ -46,6 +46,7 @@ #include "shader/prog_print.h" #include "shader/prog_statevars.h" #include "slang_typeinfo.h" +#include "slang_builtin.h" #include "slang_codegen.h" #include "slang_compile.h" #include "slang_label.h" @@ -342,109 +343,6 @@ slang_operation_identifier(slang_operation *oper, } -#define SWIZZLE_ZWWW MAKE_SWIZZLE4(SWIZZLE_Z, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W) - -/** - * Return the VERT_ATTRIB_* or FRAG_ATTRIB_* value that corresponds to - * a vertex or fragment program input variable. Return -1 if the input - * name is invalid. - * XXX return size too - */ -static GLint -_slang_input_index(const char *name, GLenum target, GLuint *swizzleOut) -{ - struct input_info { - const char *Name; - GLuint Attrib; - GLuint Swizzle; - }; - static const struct input_info vertInputs[] = { - { "gl_Vertex", VERT_ATTRIB_POS, SWIZZLE_NOOP }, - { "gl_Normal", VERT_ATTRIB_NORMAL, SWIZZLE_NOOP }, - { "gl_Color", VERT_ATTRIB_COLOR0, SWIZZLE_NOOP }, - { "gl_SecondaryColor", VERT_ATTRIB_COLOR1, SWIZZLE_NOOP }, - { "gl_FogCoord", VERT_ATTRIB_FOG, SWIZZLE_XXXX }, - { "gl_MultiTexCoord0", VERT_ATTRIB_TEX0, SWIZZLE_NOOP }, - { "gl_MultiTexCoord1", VERT_ATTRIB_TEX1, SWIZZLE_NOOP }, - { "gl_MultiTexCoord2", VERT_ATTRIB_TEX2, SWIZZLE_NOOP }, - { "gl_MultiTexCoord3", VERT_ATTRIB_TEX3, SWIZZLE_NOOP }, - { "gl_MultiTexCoord4", VERT_ATTRIB_TEX4, SWIZZLE_NOOP }, - { "gl_MultiTexCoord5", VERT_ATTRIB_TEX5, SWIZZLE_NOOP }, - { "gl_MultiTexCoord6", VERT_ATTRIB_TEX6, SWIZZLE_NOOP }, - { "gl_MultiTexCoord7", VERT_ATTRIB_TEX7, SWIZZLE_NOOP }, - { NULL, 0, SWIZZLE_NOOP } - }; - static const struct input_info fragInputs[] = { - { "gl_FragCoord", FRAG_ATTRIB_WPOS, SWIZZLE_NOOP }, - { "gl_Color", FRAG_ATTRIB_COL0, SWIZZLE_NOOP }, - { "gl_SecondaryColor", FRAG_ATTRIB_COL1, SWIZZLE_NOOP }, - { "gl_TexCoord", FRAG_ATTRIB_TEX0, SWIZZLE_NOOP }, - /* note: we're packing several quantities into the fogcoord vector */ - { "gl_FogFragCoord", FRAG_ATTRIB_FOGC, SWIZZLE_XXXX }, - { "gl_PointCoord", FRAG_ATTRIB_PNTC, SWIZZLE_XYZW }, - { "gl_FrontFacing", FRAG_ATTRIB_FACE, SWIZZLE_XXXX }, - { NULL, 0, SWIZZLE_NOOP } - }; - GLuint i; - const struct input_info *inputs - = (target == GL_VERTEX_PROGRAM_ARB) ? vertInputs : fragInputs; - - ASSERT(MAX_TEXTURE_COORD_UNITS == 8); /* if this fails, fix vertInputs above */ - - for (i = 0; inputs[i].Name; i++) { - if (strcmp(inputs[i].Name, name) == 0) { - /* found */ - *swizzleOut = inputs[i].Swizzle; - return inputs[i].Attrib; - } - } - return -1; -} - - -/** - * Return the VERT_RESULT_* or FRAG_RESULT_* value that corresponds to - * a vertex or fragment program output variable. Return -1 for an invalid - * output name. - */ -static GLint -_slang_output_index(const char *name, GLenum target) -{ - struct output_info { - const char *Name; - GLuint Attrib; - }; - static const struct output_info vertOutputs[] = { - { "gl_Position", VERT_RESULT_HPOS }, - { "gl_FrontColor", VERT_RESULT_COL0 }, - { "gl_BackColor", VERT_RESULT_BFC0 }, - { "gl_FrontSecondaryColor", VERT_RESULT_COL1 }, - { "gl_BackSecondaryColor", VERT_RESULT_BFC1 }, - { "gl_TexCoord", VERT_RESULT_TEX0 }, - { "gl_FogFragCoord", VERT_RESULT_FOGC }, - { "gl_PointSize", VERT_RESULT_PSIZ }, - { NULL, 0 } - }; - static const struct output_info fragOutputs[] = { - { "gl_FragColor", FRAG_RESULT_COLOR }, - { "gl_FragDepth", FRAG_RESULT_DEPTH }, - { "gl_FragData", FRAG_RESULT_DATA0 }, - { NULL, 0 } - }; - GLuint i; - const struct output_info *outputs - = (target == GL_VERTEX_PROGRAM_ARB) ? vertOutputs : fragOutputs; - - for (i = 0; outputs[i].Name; i++) { - if (strcmp(outputs[i].Name, name) == 0) { - /* found */ - return outputs[i].Attrib; - } - } - return -1; -} - - /** * Called when we begin code/IR generation for a new while/do/for loop. */ diff --git a/src/mesa/shader/slang/slang_link.c b/src/mesa/shader/slang/slang_link.c index 6c19fc895b3..c5c94a65a24 100644 --- a/src/mesa/shader/slang/slang_link.c +++ b/src/mesa/shader/slang/slang_link.c @@ -40,6 +40,7 @@ #include "shader/prog_statevars.h" #include "shader/prog_uniform.h" #include "shader/shader_api.h" +#include "slang_builtin.h" #include "slang_link.h" @@ -327,6 +328,7 @@ _slang_resolve_attributes(struct gl_shader_program *shProg, GLint attribMap[MAX_VERTEX_GENERIC_ATTRIBS]; GLuint i, j; GLbitfield usedAttributes; /* generics only, not legacy attributes */ + GLbitfield inputsRead = 0x0; assert(origProg != linkedProg); assert(origProg->Target == GL_VERTEX_PROGRAM_ARB); @@ -370,6 +372,10 @@ _slang_resolve_attributes(struct gl_shader_program *shProg, for (i = 0; i < linkedProg->NumInstructions; i++) { struct prog_instruction *inst = linkedProg->Instructions + i; for (j = 0; j < 3; j++) { + if (inst->SrcReg[j].File == PROGRAM_INPUT) { + inputsRead |= (1 << inst->SrcReg[j].Index); + } + if (inst->SrcReg[j].File == PROGRAM_INPUT && inst->SrcReg[j].Index >= VERT_ATTRIB_GENERIC0) { /* @@ -431,6 +437,20 @@ _slang_resolve_attributes(struct gl_shader_program *shProg, } } + /* Handle pre-defined attributes here (gl_Vertex, gl_Normal, etc). + * When the user queries the active attributes we need to include both + * the user-defined attributes and the built-in ones. + */ + for (i = VERT_ATTRIB_POS; i < VERT_ATTRIB_GENERIC0; i++) { + if (inputsRead & (1 << i)) { + _mesa_add_attribute(linkedProg->Attributes, + _slang_vert_attrib_name(i), + 1, /* size */ + _slang_vert_attrib_type(i), + -1 /* attrib/input */); + } + } + return GL_TRUE; } diff --git a/src/mesa/sources.mak b/src/mesa/sources.mak index b4b36368cfb..d1c109c2c59 100644 --- a/src/mesa/sources.mak +++ b/src/mesa/sources.mak @@ -303,7 +303,8 @@ SPARC_API = \ sparc/glapi_sparc.S COMMON_DRIVER_SOURCES = \ - drivers/common/driverfuncs.c + drivers/common/driverfuncs.c \ + drivers/common/meta.c # Sources for building non-Gallium drivers diff --git a/src/mesa/state_tracker/st_cb_blit.c b/src/mesa/state_tracker/st_cb_blit.c index 28526234720..c741940bcf4 100644 --- a/src/mesa/state_tracker/st_cb_blit.c +++ b/src/mesa/state_tracker/st_cb_blit.c @@ -69,34 +69,107 @@ st_BlitFramebuffer(GLcontext *ctx, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter) { + const GLbitfield depthStencil = (GL_DEPTH_BUFFER_BIT | + GL_STENCIL_BUFFER_BIT); struct st_context *st = ctx->st; - const uint pFilter = ((filter == GL_NEAREST) ? PIPE_TEX_MIPFILTER_NEAREST : PIPE_TEX_MIPFILTER_LINEAR); + struct gl_framebuffer *readFB = ctx->ReadBuffer; + struct gl_framebuffer *drawFB = ctx->DrawBuffer; + + if (!_mesa_clip_blit(ctx, &srcX0, &srcY0, &srcX1, &srcY1, + &dstX0, &dstY0, &dstX1, &dstY1)) { + return; /* nothing to draw/blit */ + } + + if (st_fb_orientation(drawFB) == Y_0_TOP) { + /* invert Y for dest */ + dstY0 = drawFB->Height - dstY0; + dstY1 = drawFB->Height - dstY1; + } + + if (st_fb_orientation(readFB) == Y_0_TOP) { + /* invert Y for src */ + srcY0 = readFB->Height - srcY0; + srcY1 = readFB->Height - srcY1; + } + + if (srcY0 > srcY1 && dstY0 > dstY1) { + /* Both src and dst are upside down. Swap Y to make it + * right-side up to increase odds of using a fast path. + * Recall that all Gallium raster coords have Y=0=top. + */ + GLint tmp; + tmp = srcY0; + srcY0 = srcY1; + srcY1 = tmp; + tmp = dstY0; + dstY0 = dstY1; + dstY1 = tmp; + } if (mask & GL_COLOR_BUFFER_BIT) { struct st_renderbuffer *srcRb = - st_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer); + st_renderbuffer(readFB->_ColorReadBuffer); struct st_renderbuffer *dstRb = - st_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]); + st_renderbuffer(drawFB->_ColorDrawBuffers[0]); struct pipe_surface *srcSurf = srcRb->surface; struct pipe_surface *dstSurf = dstRb->surface; - if (st_fb_orientation(ctx->DrawBuffer) == Y_0_TOP) { - /* invert Y */ - srcY0 = srcRb->Base.Height - srcY0; - srcY1 = srcRb->Base.Height - srcY1; - - dstY0 = dstRb->Base.Height - dstY0; - dstY1 = dstRb->Base.Height - dstY1; - } - util_blit_pixels(st->blit, srcSurf, srcX0, srcY0, srcX1, srcY1, dstSurf, dstX0, dstY0, dstX1, dstY1, 0.0, pFilter); + } + if (mask & depthStencil) { + /* depth and/or stencil blit */ + + /* get src/dst depth surfaces */ + struct st_renderbuffer *srcDepthRb = + st_renderbuffer(readFB->Attachment[BUFFER_DEPTH].Renderbuffer); + struct st_renderbuffer *dstDepthRb = + st_renderbuffer(drawFB->Attachment[BUFFER_DEPTH].Renderbuffer); + struct pipe_surface *srcDepthSurf = + srcDepthRb ? srcDepthRb->surface : NULL; + struct pipe_surface *dstDepthSurf = + dstDepthRb ? dstDepthRb->surface : NULL; + + /* get src/dst stencil surfaces */ + struct st_renderbuffer *srcStencilRb = + st_renderbuffer(readFB->Attachment[BUFFER_STENCIL].Renderbuffer); + struct st_renderbuffer *dstStencilRb = + st_renderbuffer(drawFB->Attachment[BUFFER_STENCIL].Renderbuffer); + struct pipe_surface *srcStencilSurf = + srcStencilRb ? srcStencilRb->surface : NULL; + struct pipe_surface *dstStencilSurf = + dstStencilRb ? dstStencilRb->surface : NULL; + + if ((mask & depthStencil) == depthStencil && + srcDepthSurf == srcStencilSurf && + dstDepthSurf == dstStencilSurf) { + /* Blitting depth and stencil values between combined + * depth/stencil buffers. This is the ideal case for such buffers. + */ + util_blit_pixels(st->blit, + srcDepthSurf, srcX0, srcY0, srcX1, srcY1, + dstDepthSurf, dstX0, dstY0, dstX1, dstY1, + 0.0, pFilter); + } + else { + /* blitting depth and stencil separately */ + + if (mask & GL_DEPTH_BUFFER_BIT) { + /* blit Z only */ + _mesa_problem(ctx, "st_BlitFramebuffer(DEPTH) not completed"); + } + + if (mask & GL_STENCIL_BUFFER_BIT) { + /* blit stencil only */ + _mesa_problem(ctx, "st_BlitFramebuffer(STENCIL) not completed"); + } + } } } diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c index 668acbccb88..8a8c99f7e17 100644 --- a/src/mesa/state_tracker/st_cb_clear.c +++ b/src/mesa/state_tracker/st_cb_clear.c @@ -280,7 +280,11 @@ clear_with_quad(GLcontext *ctx, static INLINE GLboolean check_clear_color_with_quad(GLcontext *ctx, struct gl_renderbuffer *rb) { - if (ctx->Scissor.Enabled) + if (ctx->Scissor.Enabled && + (ctx->Scissor.X != 0 || + ctx->Scissor.Y != 0 || + ctx->Scissor.Width < rb->Width || + ctx->Scissor.Height < rb->Height)) return TRUE; if (!ctx->Color.ColorMask[0] || @@ -300,7 +304,11 @@ check_clear_depth_stencil_with_quad(GLcontext *ctx, struct gl_renderbuffer *rb) GLboolean maskStencil = (ctx->Stencil.WriteMask[0] & stencilMax) != stencilMax; - if (ctx->Scissor.Enabled) + if (ctx->Scissor.Enabled && + (ctx->Scissor.X != 0 || + ctx->Scissor.Y != 0 || + ctx->Scissor.Width < rb->Width || + ctx->Scissor.Height < rb->Height)) return TRUE; if (maskStencil) @@ -319,7 +327,11 @@ check_clear_depth_with_quad(GLcontext *ctx, struct gl_renderbuffer *rb) const struct st_renderbuffer *strb = st_renderbuffer(rb); const GLboolean isDS = pf_is_depth_and_stencil(strb->surface->format); - if (ctx->Scissor.Enabled) + if (ctx->Scissor.Enabled && + (ctx->Scissor.X != 0 || + ctx->Scissor.Y != 0 || + ctx->Scissor.Width < rb->Width || + ctx->Scissor.Height < rb->Height)) return TRUE; if (isDS && @@ -345,7 +357,11 @@ check_clear_stencil_with_quad(GLcontext *ctx, struct gl_renderbuffer *rb) if (maskStencil) return TRUE; - if (ctx->Scissor.Enabled) + if (ctx->Scissor.Enabled && + (ctx->Scissor.X != 0 || + ctx->Scissor.Y != 0 || + ctx->Scissor.Width < rb->Width || + ctx->Scissor.Height < rb->Height)) return TRUE; /* This is correct, but it is necessary to look at the depth clear diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index 8b5094a04f3..d19a88fa7c1 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -60,6 +60,7 @@ #include "pipe/p_inlines.h" #include "util/u_tile.h" #include "util/u_draw_quad.h" +#include "util/u_math.h" #include "shader/prog_instruction.h" #include "cso_cache/cso_context.h" @@ -341,6 +342,7 @@ make_texture(struct st_context *st, enum pipe_format pipeFormat; GLuint cpp; GLenum baseFormat; + int ptw, pth; baseFormat = _mesa_base_format(format); @@ -355,7 +357,28 @@ make_texture(struct st_context *st, if (!pixels) return NULL; - pt = st_texture_create(st, PIPE_TEXTURE_2D, pipeFormat, 0, width, height, 1, + /* Need to use POT texture? */ + ptw = width; + pth = height; + if (!screen->get_param(screen, PIPE_CAP_NPOT_TEXTURES)) { + int l2pt, maxSize; + + l2pt = util_logbase2(width); + if (1<<l2pt != width) { + ptw = 1<<(l2pt+1); + } + l2pt = util_logbase2(height); + if (1<<l2pt != height) { + pth = 1<<(l2pt+1); + } + + /* Check against maximum texture size */ + maxSize = 1 << (pipe->screen->get_param(pipe->screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS) - 1); + assert(ptw <= maxSize); + assert(pth <= maxSize); + } + + pt = st_texture_create(st, PIPE_TEXTURE_2D, pipeFormat, 0, ptw, pth, 1, PIPE_TEXTURE_USAGE_SAMPLER); if (!pt) { _mesa_unmap_drawpix_pbo(ctx, unpack); @@ -420,7 +443,7 @@ make_texture(struct st_context *st, static void draw_quad(GLcontext *ctx, GLfloat x0, GLfloat y0, GLfloat z, GLfloat x1, GLfloat y1, const GLfloat *color, - GLboolean invertTex) + GLboolean invertTex, GLfloat maxXcoord, GLfloat maxYcoord) { struct st_context *st = ctx->st; struct pipe_context *pipe = ctx->st->pipe; @@ -435,8 +458,9 @@ draw_quad(GLcontext *ctx, GLfloat x0, GLfloat y0, GLfloat z, const GLfloat clip_y0 = y0 / fb_height * 2.0f - 1.0f; const GLfloat clip_x1 = x1 / fb_width * 2.0f - 1.0f; const GLfloat clip_y1 = y1 / fb_height * 2.0f - 1.0f; - const GLfloat sLeft = 0.0f, sRight = 1.0f; - const GLfloat tTop = invertTex, tBot = 1.0f - tTop; + const GLfloat sLeft = 0.0f, sRight = maxXcoord; + const GLfloat tTop = invertTex ? maxYcoord : 0.0f; + const GLfloat tBot = invertTex ? 0.0f : maxYcoord; GLuint tex, i; /* upper-left */ @@ -608,7 +632,9 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, y0 = (GLfloat) y; y1 = y + height * ctx->Pixel.ZoomY; - draw_quad(ctx, x0, y0, z, x1, y1, color, invertTex); + draw_quad(ctx, x0, y0, z, x1, y1, color, invertTex, + (GLfloat) width / pt->width[0], + (GLfloat) height / pt->height[0]); /* restore state */ cso_restore_rasterizer(cso); @@ -648,7 +674,7 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y, usage = PIPE_TRANSFER_READ_WRITE; else usage = PIPE_TRANSFER_WRITE; - + pt = st_cond_flush_get_tex_transfer(st_context(ctx), strb->texture, 0, 0, 0, usage, x, y, width, height); @@ -778,8 +804,8 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, return; } - _mesa_set_vp_override( ctx, TRUE ); - _mesa_update_state( ctx ); + /* Mesa state should be up to date by now */ + assert(ctx->NewState == 0x0); st_validate_state(st); @@ -807,8 +833,6 @@ st_DrawPixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, pipe_texture_reference(&pt, NULL); } } - - _mesa_set_vp_override( ctx, FALSE ); } @@ -841,7 +865,7 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, usage = PIPE_TRANSFER_READ_WRITE; else usage = PIPE_TRANSFER_WRITE; - + ptDraw = st_cond_flush_get_tex_transfer(st_context(ctx), rbDraw->texture, 0, 0, 0, usage, dstx, dsty, @@ -849,7 +873,7 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, assert(ptDraw->block.width == 1); assert(ptDraw->block.height == 1); - + /* map the stencil buffer */ drawMap = screen->transfer_map(screen, ptDraw); @@ -923,6 +947,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, struct pipe_texture *pt; GLfloat *color; enum pipe_format srcFormat, texFormat; + int ptw, pth; pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL); @@ -1004,13 +1029,34 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, height -= -srcy; srcy = 0; } - + if (height < 0) return; } + /* Need to use POT texture? */ + ptw = width; + pth = height; + if (!screen->get_param(screen, PIPE_CAP_NPOT_TEXTURES)) { + int l2pt, maxSize; + + l2pt = util_logbase2(width); + if (1<<l2pt != width) { + ptw = 1<<(l2pt+1); + } + l2pt = util_logbase2(height); + if (1<<l2pt != height) { + pth = 1<<(l2pt+1); + } + + /* Check against maximum texture size */ + maxSize = 1 << (pipe->screen->get_param(pipe->screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS) - 1); + assert(ptw <= maxSize); + assert(pth <= maxSize); + } + pt = st_texture_create(ctx->st, PIPE_TEXTURE_2D, texFormat, 0, - width, height, 1, + ptw, pth, 1, PIPE_TEXTURE_USAGE_SAMPLER); if (!pt) return; diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c index ecdb988033c..a96602878e4 100644 --- a/src/mesa/state_tracker/st_cb_fbo.c +++ b/src/mesa/state_tracker/st_cb_fbo.c @@ -446,6 +446,35 @@ st_finish_render_texture(GLcontext *ctx, /** + * Validate a renderbuffer attachment for a particular usage. + */ + +static GLboolean +st_validate_attachment(struct pipe_screen *screen, + const struct gl_renderbuffer_attachment *att, + GLuint usage) +{ + const struct st_texture_object *stObj = + st_texture_object(att->Texture); + + /** + * Only validate texture attachments for now, since + * st_renderbuffer_alloc_storage makes sure that + * the format is supported. + */ + + if (att->Type != GL_TEXTURE) + return GL_TRUE; + + if (!stObj) + return GL_FALSE; + + return screen->is_format_supported(screen, stObj->pt->format, + PIPE_TEXTURE_2D, + usage, 0); +} + +/** * Check that the framebuffer configuration is valid in terms of what * the driver can support. * @@ -454,13 +483,37 @@ st_finish_render_texture(GLcontext *ctx, static void st_validate_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb) { + struct pipe_screen *screen = ctx->st->pipe->screen; const struct gl_renderbuffer *depthRb = fb->Attachment[BUFFER_DEPTH].Renderbuffer; const struct gl_renderbuffer *stencilRb = fb->Attachment[BUFFER_STENCIL].Renderbuffer; + GLuint i; if (stencilRb && depthRb && stencilRb != depthRb) { fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; + return; + } + + if (!st_validate_attachment(screen, + &fb->Attachment[BUFFER_DEPTH], + PIPE_TEXTURE_USAGE_DEPTH_STENCIL)) { + fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; + return; + } + if (!st_validate_attachment(screen, + &fb->Attachment[BUFFER_STENCIL], + PIPE_TEXTURE_USAGE_DEPTH_STENCIL)) { + fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; + return; + } + for (i = 0; i < ctx->Const.MaxColorAttachments; i++) { + if (!st_validate_attachment(screen, + &fb->Attachment[BUFFER_COLOR0 + i], + PIPE_TEXTURE_USAGE_RENDER_TARGET)) { + fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; + return; + } } } diff --git a/src/mesa/state_tracker/st_cb_flush.c b/src/mesa/state_tracker/st_cb_flush.c index 8ceeeabcd37..1329f807bc9 100644 --- a/src/mesa/state_tracker/st_cb_flush.c +++ b/src/mesa/state_tracker/st_cb_flush.c @@ -161,4 +161,16 @@ void st_init_flush_functions(struct dd_function_table *functions) { functions->Flush = st_glFlush; functions->Finish = st_glFinish; + + /* Windows opengl32.dll calls glFinish prior to every swapbuffers. + * This is unnecessary and degrades performance. Luckily we have some + * scope to work around this, as the externally-visible behaviour of + * Finish() is identical to Flush() in all cases - no differences in + * rendering or ReadPixels are visible if we opt not to wait here. + * + * Only set this up on windows to avoid suprise elsewhere. + */ +#ifdef PIPE_OS_WINDOWS + functions->Finish = st_glFlush; +#endif } diff --git a/src/mesa/state_tracker/st_cb_rasterpos.c b/src/mesa/state_tracker/st_cb_rasterpos.c index 3bcccd0df46..d82b2a2035f 100644 --- a/src/mesa/state_tracker/st_cb_rasterpos.c +++ b/src/mesa/state_tracker/st_cb_rasterpos.c @@ -251,7 +251,7 @@ st_RasterPos(GLcontext *ctx, const GLfloat v[4]) rs->array[0].Ptr = (GLubyte *) v; /* draw the point */ - st_feedback_draw_vbo(ctx, rs->arrays, &rs->prim, 1, NULL, 0, 1); + st_feedback_draw_vbo(ctx, rs->arrays, &rs->prim, 1, NULL, GL_TRUE, 0, 1); } diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index ee71c012c64..90a059ca69a 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -59,6 +59,7 @@ #include "util/u_tile.h" #include "util/u_blit.h" #include "util/u_surface.h" +#include "util/u_math.h" #define DBG if (0) printf @@ -237,16 +238,6 @@ do_memcpy(void *dest, const void *src, size_t n) } -static INLINE unsigned -logbase2(unsigned n) -{ - unsigned log2 = 0; - while (n >>= 1) - ++log2; - return log2; -} - - /** * Return default texture usage bitmask for the given texture format. */ @@ -340,9 +331,9 @@ guess_and_alloc_texture(struct st_context *st, lastLevel = firstLevel; } else { - GLuint l2width = logbase2(width); - GLuint l2height = logbase2(height); - GLuint l2depth = logbase2(depth); + GLuint l2width = util_logbase2(width); + GLuint l2height = util_logbase2(height); + GLuint l2depth = util_logbase2(depth); lastLevel = firstLevel + MAX2(MAX2(l2width, l2height), l2depth); } @@ -532,6 +523,12 @@ st_TexImage(GLcontext * ctx, DBG("%s target %s level %d %dx%dx%d border %d\n", __FUNCTION__, _mesa_lookup_enum_by_nr(target), level, width, height, depth, border); + /* switch to "normal" */ + if (stObj->surface_based) { + _mesa_clear_texture_object(ctx, texObj); + stObj->surface_based = GL_FALSE; + } + /* gallium does not support texture borders, strip it off */ if (border) { strip_texture_border(border, &width, &height, &depth, unpack, &unpackNB); diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c index 914a507bef6..503a5f34a3f 100644 --- a/src/mesa/state_tracker/st_draw.c +++ b/src/mesa/state_tracker/st_draw.c @@ -533,6 +533,7 @@ st_draw_vbo(GLcontext *ctx, const struct _mesa_prim *prims, GLuint nr_prims, const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, GLuint min_index, GLuint max_index) { @@ -545,6 +546,10 @@ st_draw_vbo(GLcontext *ctx, unsigned num_vbuffers, num_velements; GLboolean userSpace; + /* Gallium probably doesn't want this in some cases. */ + if (!index_bounds_valid) + vbo_get_minmax_index(ctx, prims, ib, &min_index, &max_index); + /* sanity check for pointer arithmetic below */ assert(sizeof(arrays[0]->Ptr[0]) == 1); diff --git a/src/mesa/state_tracker/st_draw.h b/src/mesa/state_tracker/st_draw.h index dcfe7e15361..3e0face656b 100644 --- a/src/mesa/state_tracker/st_draw.h +++ b/src/mesa/state_tracker/st_draw.h @@ -47,6 +47,7 @@ st_draw_vbo(GLcontext *ctx, const struct _mesa_prim *prims, GLuint nr_prims, const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, GLuint min_index, GLuint max_index); @@ -56,6 +57,7 @@ st_feedback_draw_vbo(GLcontext *ctx, const struct _mesa_prim *prims, GLuint nr_prims, const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, GLuint min_index, GLuint max_index); diff --git a/src/mesa/state_tracker/st_draw_feedback.c b/src/mesa/state_tracker/st_draw_feedback.c index 2712c131c0d..b2d682ef640 100644 --- a/src/mesa/state_tracker/st_draw_feedback.c +++ b/src/mesa/state_tracker/st_draw_feedback.c @@ -96,6 +96,7 @@ st_feedback_draw_vbo(GLcontext *ctx, const struct _mesa_prim *prims, GLuint nr_prims, const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, GLuint min_index, GLuint max_index) { @@ -114,6 +115,9 @@ st_feedback_draw_vbo(GLcontext *ctx, st_validate_state(ctx->st); + if (!index_bounds_valid) + vbo_get_minmax_index(ctx, prims, ib, &min_index, &max_index); + /* must get these after state validation! */ vp = ctx->st->vp; vs = &st->vp->state; diff --git a/src/mesa/state_tracker/st_extensions.c b/src/mesa/state_tracker/st_extensions.c index 8a958e8bd88..3f835d38dd5 100644 --- a/src/mesa/state_tracker/st_extensions.c +++ b/src/mesa/state_tracker/st_extensions.c @@ -233,7 +233,9 @@ void st_init_extensions(struct st_context *st) if (screen->get_param(screen, PIPE_CAP_POINT_SPRITE)) { ctx->Extensions.ARB_point_sprite = GL_TRUE; - ctx->Extensions.NV_point_sprite = GL_TRUE; + /* GL_NV_point_sprite is not supported by gallium because we don't + * support the GL_POINT_SPRITE_R_MODE_NV option. + */ } if (screen->get_param(screen, PIPE_CAP_OCCLUSION_QUERY)) { diff --git a/src/mesa/state_tracker/st_format.c b/src/mesa/state_tracker/st_format.c index b243c249e37..dcb90a31076 100644 --- a/src/mesa/state_tracker/st_format.c +++ b/src/mesa/state_tracker/st_format.c @@ -435,10 +435,10 @@ st_choose_format(struct pipe_context *pipe, GLenum internalFormat, case GL_RGB5: case GL_RGB4: case GL_R3_G3_B2: - if (screen->is_format_supported( screen, PIPE_FORMAT_A1R5G5B5_UNORM, target, tex_usage, geom_flags )) - return PIPE_FORMAT_A1R5G5B5_UNORM; if (screen->is_format_supported( screen, PIPE_FORMAT_R5G6B5_UNORM, target, tex_usage, geom_flags )) return PIPE_FORMAT_R5G6B5_UNORM; + if (screen->is_format_supported( screen, PIPE_FORMAT_A1R5G5B5_UNORM, target, tex_usage, geom_flags )) + return PIPE_FORMAT_A1R5G5B5_UNORM; return default_rgba_format( screen, target, tex_usage, geom_flags ); case GL_ALPHA: diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c index 210ca82c42e..2ab12d3cf3f 100644 --- a/src/mesa/state_tracker/st_mesa_to_tgsi.c +++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c @@ -137,12 +137,6 @@ map_texture_target( GLuint textarget, GLboolean shadow ) { -#if 1 - /* XXX remove this line after we've checked that the rest of gallium - * can handle the TGSI_TEXTURE_SHADOWx tokens. - */ - shadow = GL_FALSE; -#endif switch( textarget ) { case TEXTURE_1D_INDEX: if (shadow) @@ -353,7 +347,7 @@ compile_instruction( fullinst->Instruction.Opcode = TGSI_OPCODE_ADD; break; case OPCODE_BGNLOOP: - fullinst->Instruction.Opcode = TGSI_OPCODE_BGNLOOP2; + fullinst->Instruction.Opcode = TGSI_OPCODE_BGNLOOP; fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size; break; case OPCODE_BGNSUB: @@ -411,7 +405,7 @@ compile_instruction( fullinst->Instruction.Opcode = TGSI_OPCODE_ENDIF; break; case OPCODE_ENDLOOP: - fullinst->Instruction.Opcode = TGSI_OPCODE_ENDLOOP2; + fullinst->Instruction.Opcode = TGSI_OPCODE_ENDLOOP; fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size; break; case OPCODE_ENDSUB: diff --git a/src/mesa/state_tracker/st_public.h b/src/mesa/state_tracker/st_public.h index 04d3a3d7c2c..a5fdac32d1f 100644 --- a/src/mesa/state_tracker/st_public.h +++ b/src/mesa/state_tracker/st_public.h @@ -107,7 +107,9 @@ void st_swapbuffers(struct st_framebuffer *stfb, struct pipe_surface **front_left, struct pipe_surface **front_right); -int st_set_teximage(struct pipe_texture *pt, int target); +int st_bind_texture_surface(struct pipe_surface *ps, int target, int level, + enum pipe_format format); +int st_unbind_texture_surface(struct pipe_surface *ps, int target, int level); /** Redirect rendering into stfb's surface to a texture image */ int st_bind_teximage(struct st_framebuffer *stfb, uint surfIndex, diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c index 10faa633ea8..d58803991a5 100644 --- a/src/mesa/state_tracker/st_texture.c +++ b/src/mesa/state_tracker/st_texture.c @@ -32,6 +32,7 @@ #include "st_cb_fbo.h" #include "st_inlines.h" #include "main/enums.h" +#include "main/texobj.h" #include "main/teximage.h" #include "main/texstore.h" @@ -353,25 +354,95 @@ st_texture_image_copy(struct pipe_context *pipe, } } -/** Bind a pipe surface for use as a texture image */ + +/** + * Bind a pipe surface to a texture object. After the call, + * the texture object is marked dirty and will be (re-)validated. + * + * If this is the first surface bound, the texture object is said to + * switch from normal to surface based. It will be cleared first in + * this case. + * + * \param ps pipe surface to be unbound + * \param target texture target + * \param level image level + * \param format internal format of the texture + */ int -st_set_teximage(struct pipe_texture *pt, int target) +st_bind_texture_surface(struct pipe_surface *ps, int target, int level, + enum pipe_format format) { GET_CURRENT_CONTEXT(ctx); const GLuint unit = ctx->Texture.CurrentUnit; struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit]; struct gl_texture_object *texObj; struct gl_texture_image *texImage; + struct st_texture_object *stObj; struct st_texture_image *stImage; - int internalFormat; + GLenum internalFormat; - switch (pt->format) { - case PIPE_FORMAT_A8R8G8B8_UNORM: - internalFormat = GL_RGBA8; + switch (target) { + case ST_TEXTURE_2D: + target = GL_TEXTURE_2D; + break; + case ST_TEXTURE_RECT: + target = GL_TEXTURE_RECTANGLE_ARB; break; default: return 0; - }; + } + + /* map pipe format to base format for now */ + if (pf_get_component_bits(format, PIPE_FORMAT_COMP_A) > 0) + internalFormat = GL_RGBA; + else + internalFormat = GL_RGB; + + texObj = _mesa_select_tex_object(ctx, texUnit, target); + _mesa_lock_texture(ctx, texObj); + + stObj = st_texture_object(texObj); + /* switch to surface based */ + if (!stObj->surface_based) { + _mesa_clear_texture_object(ctx, texObj); + stObj->surface_based = GL_TRUE; + } + + texImage = _mesa_get_tex_image(ctx, texObj, target, level); + stImage = st_texture_image(texImage); + + _mesa_init_teximage_fields(ctx, target, texImage, + ps->width, ps->height, 1, 0, internalFormat); + texImage->TexFormat = st_ChooseTextureFormat(ctx, internalFormat, + GL_RGBA, GL_UNSIGNED_BYTE); + _mesa_set_fetch_functions(texImage, 2); + pipe_texture_reference(&stImage->pt, ps->texture); + + _mesa_dirty_texobj(ctx, texObj, GL_TRUE); + _mesa_unlock_texture(ctx, texObj); + + return 1; +} + + +/** + * Unbind a pipe surface from a texture object. After the call, + * the texture object is marked dirty and will be (re-)validated. + * + * \param ps pipe surface to be unbound + * \param target texture target + * \param level image level + */ +int +st_unbind_texture_surface(struct pipe_surface *ps, int target, int level) +{ + GET_CURRENT_CONTEXT(ctx); + const GLuint unit = ctx->Texture.CurrentUnit; + struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit]; + struct gl_texture_object *texObj; + struct gl_texture_image *texImage; + struct st_texture_object *stObj; + struct st_texture_image *stImage; switch (target) { case ST_TEXTURE_2D: @@ -385,21 +456,28 @@ st_set_teximage(struct pipe_texture *pt, int target) } texObj = _mesa_select_tex_object(ctx, texUnit, target); - texImage = _mesa_get_tex_image(ctx, texObj, target, 0); + + _mesa_lock_texture(ctx, texObj); + + texImage = _mesa_get_tex_image(ctx, texObj, target, level); + stObj = st_texture_object(texObj); stImage = st_texture_image(texImage); - - _mesa_init_teximage_fields(ctx, GL_TEXTURE_2D, texImage, pt->width[0], - pt->height[0], 1, 0, internalFormat); - texImage->TexFormat = st_ChooseTextureFormat(ctx, internalFormat, GL_RGBA, - GL_UNSIGNED_BYTE); - _mesa_set_fetch_functions(texImage, 2); + /* Make sure the pipe surface is still bound. The texture object is still + * considered surface based even if this is the last bound surface. */ + if (stImage->pt == ps->texture) { + pipe_texture_reference(&stImage->pt, NULL); + _mesa_clear_texture_image(ctx, texImage); - pipe_texture_reference(&stImage->pt, pt); + _mesa_dirty_texobj(ctx, texObj, GL_TRUE); + } + _mesa_unlock_texture(ctx, texObj); + return 1; } + /** Redirect rendering into stfb's surface to a texture image */ int st_bind_teximage(struct st_framebuffer *stfb, uint surfIndex, diff --git a/src/mesa/state_tracker/st_texture.h b/src/mesa/state_tracker/st_texture.h index b9d447cb56e..60868ce0673 100644 --- a/src/mesa/state_tracker/st_texture.h +++ b/src/mesa/state_tracker/st_texture.h @@ -69,6 +69,11 @@ struct st_texture_object struct pipe_texture *pt; GLboolean teximage_realloc; + + /* True if there is/was a surface bound to this texture object. It helps + * track whether the texture object is surface based or not. + */ + GLboolean surface_based; }; diff --git a/src/mesa/swrast/s_blit.c b/src/mesa/swrast/s_blit.c index 0e32cb8f653..8303e4debc7 100644 --- a/src/mesa/swrast/s_blit.c +++ b/src/mesa/swrast/s_blit.c @@ -24,6 +24,7 @@ #include "main/glheader.h" +#include "main/image.h" #include "main/macros.h" #include "s_context.h" @@ -104,7 +105,7 @@ static void blit_nearest(GLcontext *ctx, GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, - GLenum buffer) + GLbitfield buffer) { struct gl_renderbuffer *readRb, *drawRb; @@ -456,7 +457,7 @@ static void simple_blit(GLcontext *ctx, GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, - GLenum buffer) + GLbitfield buffer) { struct gl_renderbuffer *readRb, *drawRb; const GLint width = srcX1 - srcX0; @@ -551,184 +552,6 @@ simple_blit(GLcontext *ctx, /** - * Clip dst coords against Xmax (or Ymax). - */ -static INLINE void -clip_right_or_top(GLint *srcX0, GLint *srcX1, - GLint *dstX0, GLint *dstX1, - GLint maxValue) -{ - GLfloat t, bias; - - if (*dstX1 > maxValue) { - /* X1 outside right edge */ - ASSERT(*dstX0 < maxValue); /* X0 should be inside right edge */ - t = (GLfloat) (maxValue - *dstX0) / (GLfloat) (*dstX1 - *dstX0); - /* chop off [t, 1] part */ - ASSERT(t >= 0.0 && t <= 1.0); - *dstX1 = maxValue; - bias = (*srcX0 < *srcX1) ? 0.5 : -0.5; - *srcX1 = *srcX0 + (GLint) (t * (*srcX1 - *srcX0) + bias); - } - else if (*dstX0 > maxValue) { - /* X0 outside right edge */ - ASSERT(*dstX1 < maxValue); /* X1 should be inside right edge */ - t = (GLfloat) (maxValue - *dstX1) / (GLfloat) (*dstX0 - *dstX1); - /* chop off [t, 1] part */ - ASSERT(t >= 0.0 && t <= 1.0); - *dstX0 = maxValue; - bias = (*srcX0 < *srcX1) ? -0.5 : 0.5; - *srcX0 = *srcX1 + (GLint) (t * (*srcX0 - *srcX1) + bias); - } -} - - -/** - * Clip dst coords against Xmin (or Ymin). - */ -static INLINE void -clip_left_or_bottom(GLint *srcX0, GLint *srcX1, - GLint *dstX0, GLint *dstX1, - GLint minValue) -{ - GLfloat t, bias; - - if (*dstX0 < minValue) { - /* X0 outside left edge */ - ASSERT(*dstX1 > minValue); /* X1 should be inside left edge */ - t = (GLfloat) (minValue - *dstX0) / (GLfloat) (*dstX1 - *dstX0); - /* chop off [0, t] part */ - ASSERT(t >= 0.0 && t <= 1.0); - *dstX0 = minValue; - bias = (*srcX0 < *srcX1) ? 0.5 : -0.5; /* flipped??? */ - *srcX0 = *srcX0 + (GLint) (t * (*srcX1 - *srcX0) + bias); - } - else if (*dstX1 < minValue) { - /* X1 outside left edge */ - ASSERT(*dstX0 > minValue); /* X0 should be inside left edge */ - t = (GLfloat) (minValue - *dstX1) / (GLfloat) (*dstX0 - *dstX1); - /* chop off [0, t] part */ - ASSERT(t >= 0.0 && t <= 1.0); - *dstX1 = minValue; - bias = (*srcX0 < *srcX1) ? 0.5 : -0.5; - *srcX1 = *srcX1 + (GLint) (t * (*srcX0 - *srcX1) + bias); - } -} - - -/** - * Do clipping of blit src/dest rectangles. - * The dest rect is clipped against both the buffer bounds and scissor bounds. - * The src rect is just clipped against the buffer bounds. - * - * When either the src or dest rect is clipped, the other is also clipped - * proportionately! - * - * Note that X0 need not be less than X1 (same for Y) for either the source - * and dest rects. That makes the clipping a little trickier. - * - * \return GL_TRUE if anything is left to draw, GL_FALSE if totally clipped - */ -static GLboolean -clip_blit(GLcontext *ctx, - GLint *srcX0, GLint *srcY0, GLint *srcX1, GLint *srcY1, - GLint *dstX0, GLint *dstY0, GLint *dstX1, GLint *dstY1) -{ - const GLint srcXmin = 0; - const GLint srcXmax = ctx->ReadBuffer->Width; - const GLint srcYmin = 0; - const GLint srcYmax = ctx->ReadBuffer->Height; - - /* these include scissor bounds */ - const GLint dstXmin = ctx->DrawBuffer->_Xmin; - const GLint dstXmax = ctx->DrawBuffer->_Xmax; - const GLint dstYmin = ctx->DrawBuffer->_Ymin; - const GLint dstYmax = ctx->DrawBuffer->_Ymax; - - /* - printf("PreClipX: src: %d .. %d dst: %d .. %d\n", - *srcX0, *srcX1, *dstX0, *dstX1); - printf("PreClipY: src: %d .. %d dst: %d .. %d\n", - *srcY0, *srcY1, *dstY0, *dstY1); - */ - - /* trivial rejection tests */ - if (*dstX0 == *dstX1) - return GL_FALSE; /* no width */ - if (*dstX0 <= dstXmin && *dstX1 <= dstXmin) - return GL_FALSE; /* totally out (left) of bounds */ - if (*dstX0 >= dstXmax && *dstX1 >= dstXmax) - return GL_FALSE; /* totally out (right) of bounds */ - - if (*dstY0 == *dstY1) - return GL_FALSE; - if (*dstY0 <= dstYmin && *dstY1 <= dstYmin) - return GL_FALSE; - if (*dstY0 >= dstYmax && *dstY1 >= dstYmax) - return GL_FALSE; - - if (*srcX0 == *srcX1) - return GL_FALSE; - if (*srcX0 <= srcXmin && *srcX1 <= srcXmin) - return GL_FALSE; - if (*srcX0 >= srcXmax && *srcX1 >= srcXmax) - return GL_FALSE; - - if (*srcY0 == *srcY1) - return GL_FALSE; - if (*srcY0 <= srcYmin && *srcY1 <= srcYmin) - return GL_FALSE; - if (*srcY0 >= srcYmax && *srcY1 >= srcYmax) - return GL_FALSE; - - /* - * dest clip - */ - clip_right_or_top(srcX0, srcX1, dstX0, dstX1, dstXmax); - clip_right_or_top(srcY0, srcY1, dstY0, dstY1, dstYmax); - clip_left_or_bottom(srcX0, srcX1, dstX0, dstX1, dstXmin); - clip_left_or_bottom(srcY0, srcY1, dstY0, dstY1, dstYmin); - - /* - * src clip (just swap src/dst values from above) - */ - clip_right_or_top(dstX0, dstX1, srcX0, srcX1, srcXmax); - clip_right_or_top(dstY0, dstY1, srcY0, srcY1, srcYmax); - clip_left_or_bottom(dstX0, dstX1, srcX0, srcX1, srcXmin); - clip_left_or_bottom(dstY0, dstY1, srcY0, srcY1, srcYmin); - - /* - printf("PostClipX: src: %d .. %d dst: %d .. %d\n", - *srcX0, *srcX1, *dstX0, *dstX1); - printf("PostClipY: src: %d .. %d dst: %d .. %d\n", - *srcY0, *srcY1, *dstY0, *dstY1); - */ - - ASSERT(*dstX0 >= dstXmin); - ASSERT(*dstX0 <= dstXmax); - ASSERT(*dstX1 >= dstXmin); - ASSERT(*dstX1 <= dstXmax); - - ASSERT(*dstY0 >= dstYmin); - ASSERT(*dstY0 <= dstYmax); - ASSERT(*dstY1 >= dstYmin); - ASSERT(*dstY1 <= dstYmax); - - ASSERT(*srcX0 >= srcXmin); - ASSERT(*srcX0 <= srcXmax); - ASSERT(*srcX1 >= srcXmin); - ASSERT(*srcX1 <= srcXmax); - - ASSERT(*srcY0 >= srcYmin); - ASSERT(*srcY0 <= srcYmax); - ASSERT(*srcY1 >= srcYmin); - ASSERT(*srcY1 <= srcYmax); - - return GL_TRUE; -} - - -/** * Software fallback for glBlitFramebufferEXT(). */ void @@ -737,7 +560,7 @@ _swrast_BlitFramebuffer(GLcontext *ctx, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter) { - static const GLint buffers[3] = { + static const GLbitfield buffers[3] = { GL_COLOR_BUFFER_BIT, GL_DEPTH_BUFFER_BIT, GL_STENCIL_BUFFER_BIT @@ -747,8 +570,8 @@ _swrast_BlitFramebuffer(GLcontext *ctx, if (!ctx->DrawBuffer->_NumColorDrawBuffers) return; - if (!clip_blit(ctx, &srcX0, &srcY0, &srcX1, &srcY1, - &dstX0, &dstY0, &dstX1, &dstY1)) { + if (!_mesa_clip_blit(ctx, &srcX0, &srcY0, &srcX1, &srcY1, + &dstX0, &dstY0, &dstX1, &dstY1)) { return; } diff --git a/src/mesa/tnl/t_context.c b/src/mesa/tnl/t_context.c index f69b1220461..f2771cde095 100644 --- a/src/mesa/tnl/t_context.c +++ b/src/mesa/tnl/t_context.c @@ -81,7 +81,7 @@ _tnl_CreateContext( GLcontext *ctx ) tnl->nr_blocks = 0; /* plug in the VBO drawing function */ - vbo_set_draw_func(ctx, _tnl_draw_prims); + vbo_set_draw_func(ctx, _tnl_vbo_draw_prims); _math_init_transformation(); _math_init_translate(); diff --git a/src/mesa/tnl/t_draw.c b/src/mesa/tnl/t_draw.c index 2ec65d53233..c64c2c20778 100644 --- a/src/mesa/tnl/t_draw.c +++ b/src/mesa/tnl/t_draw.c @@ -360,6 +360,20 @@ static void unmap_vbos( GLcontext *ctx, } +void _tnl_vbo_draw_prims(GLcontext *ctx, + const struct gl_client_array *arrays[], + const struct _mesa_prim *prim, + GLuint nr_prims, + const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, + GLuint min_index, + GLuint max_index) +{ + if (!index_bounds_valid) + vbo_get_minmax_index(ctx, prim, ib, &min_index, &max_index); + + _tnl_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index); +} /* This is the main entrypoint into the slimmed-down software tnl * module. In a regular swtnl driver, this can be plugged straight @@ -393,7 +407,7 @@ void _tnl_draw_prims( GLcontext *ctx, */ vbo_rebase_prims( ctx, arrays, prim, nr_prims, ib, min_index, max_index, - _tnl_draw_prims ); + _tnl_vbo_draw_prims ); return; } else if (max_index > max) { @@ -411,7 +425,7 @@ void _tnl_draw_prims( GLcontext *ctx, */ vbo_split_prims( ctx, arrays, prim, nr_prims, ib, 0, max_index, - _tnl_draw_prims, + _tnl_vbo_draw_prims, &limits ); } else { diff --git a/src/mesa/tnl/tnl.h b/src/mesa/tnl/tnl.h index 4d628aa9a60..9c66d3b0192 100644 --- a/src/mesa/tnl/tnl.h +++ b/src/mesa/tnl/tnl.h @@ -81,6 +81,16 @@ _tnl_draw_prims( GLcontext *ctx, GLuint min_index, GLuint max_index); +void +_tnl_vbo_draw_prims( GLcontext *ctx, + const struct gl_client_array *arrays[], + const struct _mesa_prim *prim, + GLuint nr_prims, + const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, + GLuint min_index, + GLuint max_index); + extern void _mesa_load_tracked_matrices(GLcontext *ctx); diff --git a/src/mesa/vbo/vbo.h b/src/mesa/vbo/vbo.h index 5362226c2f9..5986e93576c 100644 --- a/src/mesa/vbo/vbo.h +++ b/src/mesa/vbo/vbo.h @@ -69,6 +69,7 @@ typedef void (*vbo_draw_func)( GLcontext *ctx, const struct _mesa_prim *prims, GLuint nr_prims, const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, GLuint min_index, GLuint max_index ); @@ -112,7 +113,10 @@ void vbo_rebase_prims( GLcontext *ctx, GLuint min_index, GLuint max_index, vbo_draw_func draw ); - +void +vbo_get_minmax_index(GLcontext *ctx, const struct _mesa_prim *prim, + const struct _mesa_index_buffer *ib, + GLuint *min_index, GLuint *max_index); void vbo_use_buffer_objects(GLcontext *ctx); diff --git a/src/mesa/vbo/vbo_exec_array.c b/src/mesa/vbo/vbo_exec_array.c index f4b9b2f7443..ddf6ca1e070 100644 --- a/src/mesa/vbo/vbo_exec_array.c +++ b/src/mesa/vbo/vbo_exec_array.c @@ -41,13 +41,27 @@ /** * Compute min and max elements for glDraw[Range]Elements() calls. */ -static void -get_minmax_index(GLuint count, GLuint type, const GLvoid *indices, - GLuint *min_index, GLuint *max_index) +void +vbo_get_minmax_index(GLcontext *ctx, + const struct _mesa_prim *prim, + const struct _mesa_index_buffer *ib, + GLuint *min_index, GLuint *max_index) { GLuint i; + GLsizei count = prim->count; + const void *indices; + + if (_mesa_is_bufferobj(ib->obj)) { + const GLvoid *map = ctx->Driver.MapBuffer(ctx, + GL_ELEMENT_ARRAY_BUFFER_ARB, + GL_READ_ONLY, + ib->obj); + indices = ADD_POINTERS(map, ib->ptr); + } else { + indices = ib->ptr; + } - switch(type) { + switch (ib->type) { case GL_UNSIGNED_INT: { const GLuint *ui_indices = (const GLuint *)indices; GLuint max_ui = ui_indices[count-1]; @@ -88,6 +102,12 @@ get_minmax_index(GLuint count, GLuint type, const GLvoid *indices, assert(0); break; } + + if (_mesa_is_bufferobj(ib->obj)) { + ctx->Driver.UnmapBuffer(ctx, + GL_ELEMENT_ARRAY_BUFFER_ARB, + ib->obj); + } } @@ -101,7 +121,7 @@ check_array_data(GLcontext *ctx, struct gl_client_array *array, { if (array->Enabled) { const void *data = array->Ptr; - if (array->BufferObj->Name) { + if (_mesa_is_bufferobj(array->BufferObj)) { if (!array->BufferObj->Pointer) { /* need to map now */ array->BufferObj->Pointer = ctx->Driver.MapBuffer(ctx, @@ -146,8 +166,8 @@ static void unmap_array_buffer(GLcontext *ctx, struct gl_client_array *array) { if (array->Enabled && - array->BufferObj->Name && - array->BufferObj->Pointer) { + _mesa_is_bufferobj(array->BufferObj) && + _mesa_bufferobj_mapped(array->BufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER_ARB, array->BufferObj); @@ -166,7 +186,7 @@ check_draw_elements_data(GLcontext *ctx, GLsizei count, GLenum elemType, const void *elemMap; GLint i, k; - if (ctx->Array.ElementArrayBufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Array.ElementArrayBufferObj)) { elemMap = ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER_ARB, GL_READ_ONLY, @@ -205,7 +225,7 @@ check_draw_elements_data(GLcontext *ctx, GLsizei count, GLenum elemType, } } - if (ctx->Array.ElementArrayBufferObj->Name) { + if (_mesa_is_bufferobj(ctx->Array.ElementArrayBufferObj)) { ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER_ARB, ctx->Array.ElementArrayBufferObj); @@ -500,7 +520,7 @@ vbo_exec_DrawArrays(GLenum mode, GLint start, GLsizei count) prim[0].indexed = 0; vbo->draw_prims( ctx, exec->array.inputs, prim, 1, NULL, - start, start + count - 1 ); + GL_TRUE, start, start + count - 1 ); #if 0 print_draw_arrays(ctx, exec, mode, start, count); @@ -566,53 +586,19 @@ dump_element_buffer(GLcontext *ctx, GLenum type) ctx->Array.ElementArrayBufferObj); } - -static void GLAPIENTRY -vbo_exec_DrawRangeElements(GLenum mode, - GLuint start, GLuint end, - GLsizei count, GLenum type, const GLvoid *indices) +/* Inner support for both _mesa_DrawElements and _mesa_DrawRangeElements */ +static void +vbo_validated_drawrangeelements(GLcontext *ctx, GLenum mode, + GLboolean index_bounds_valid, + GLuint start, GLuint end, + GLsizei count, GLenum type, + const GLvoid *indices) { - GET_CURRENT_CONTEXT(ctx); struct vbo_context *vbo = vbo_context(ctx); struct vbo_exec_context *exec = &vbo->exec; struct _mesa_index_buffer ib; struct _mesa_prim prim[1]; - if (!_mesa_validate_DrawRangeElements( ctx, mode, start, end, count, - type, indices )) - return; - - if (end >= ctx->Array.ArrayObj->_MaxElement) { - /* the max element is out of bounds of one or more enabled arrays */ - _mesa_warning(ctx, "glDraw[Range]Elements(start %u, end %u, count %d, " - "type 0x%x, indices=%p)\n" - "\tindex=%u is out of bounds (max=%u) " - "Element Buffer %u (size %d)", - start, end, count, type, indices, end, - ctx->Array.ArrayObj->_MaxElement - 1, - ctx->Array.ElementArrayBufferObj->Name, - ctx->Array.ElementArrayBufferObj->Size); - - if (0) - dump_element_buffer(ctx, type); - - if (0) - _mesa_print_arrays(ctx); - return; - } - else if (0) { - _mesa_printf("glDraw[Range]Elements" - "(start %u, end %u, type 0x%x, count %d) ElemBuf %u\n", - start, end, type, count, - ctx->Array.ElementArrayBufferObj->Name); - } - -#if 0 - check_draw_elements_data(ctx, count, type, indices); -#else - (void) check_draw_elements_data; -#endif - FLUSH_CURRENT( ctx, 0 ); if (ctx->NewState) @@ -623,13 +609,13 @@ vbo_exec_DrawRangeElements(GLenum mode, return; } - bind_arrays( ctx ); - if (ctx->NewState) _mesa_update_state( ctx ); + bind_arrays( ctx ); + ib.count = count; - ib.type = type; + ib.type = type; ib.obj = ctx->Array.ElementArrayBufferObj; ib.ptr = indices; @@ -673,44 +659,68 @@ vbo_exec_DrawRangeElements(GLenum mode, * for the latter case elsewhere. */ - vbo->draw_prims( ctx, exec->array.inputs, prim, 1, &ib, start, end ); + vbo->draw_prims( ctx, exec->array.inputs, prim, 1, &ib, + index_bounds_valid, start, end ); } - static void GLAPIENTRY -vbo_exec_DrawElements(GLenum mode, GLsizei count, GLenum type, - const GLvoid *indices) +vbo_exec_DrawRangeElements(GLenum mode, + GLuint start, GLuint end, + GLsizei count, GLenum type, const GLvoid *indices) { GET_CURRENT_CONTEXT(ctx); - GLuint min_index = 0; - GLuint max_index = 0; - if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices )) + if (!_mesa_validate_DrawRangeElements( ctx, mode, start, end, count, + type, indices )) return; - if (!vbo_validate_shaders(ctx)) { - _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawElements(bad shader)"); + if (end >= ctx->Array.ArrayObj->_MaxElement) { + /* the max element is out of bounds of one or more enabled arrays */ + _mesa_warning(ctx, "glDraw[Range]Elements(start %u, end %u, count %d, " + "type 0x%x, indices=%p)\n" + "\tindex=%u is out of bounds (max=%u) " + "Element Buffer %u (size %d)", + start, end, count, type, indices, end, + ctx->Array.ArrayObj->_MaxElement - 1, + ctx->Array.ElementArrayBufferObj->Name, + ctx->Array.ElementArrayBufferObj->Size); + + if (0) + dump_element_buffer(ctx, type); + + if (0) + _mesa_print_arrays(ctx); return; } + else if (0) { + _mesa_printf("glDraw[Range]Elements" + "(start %u, end %u, type 0x%x, count %d) ElemBuf %u\n", + start, end, type, count, + ctx->Array.ElementArrayBufferObj->Name); + } - if (ctx->Array.ElementArrayBufferObj->Name) { - const GLvoid *map = ctx->Driver.MapBuffer(ctx, - GL_ELEMENT_ARRAY_BUFFER_ARB, - GL_READ_ONLY, - ctx->Array.ElementArrayBufferObj); +#if 0 + check_draw_elements_data(ctx, count, type, indices); +#else + (void) check_draw_elements_data; +#endif + + vbo_validated_drawrangeelements(ctx, mode, GL_TRUE, start, end, + count, type, indices); +} - get_minmax_index(count, type, ADD_POINTERS(map, indices), - &min_index, &max_index); - ctx->Driver.UnmapBuffer(ctx, - GL_ELEMENT_ARRAY_BUFFER_ARB, - ctx->Array.ElementArrayBufferObj); - } - else { - get_minmax_index(count, type, indices, &min_index, &max_index); - } +static void GLAPIENTRY +vbo_exec_DrawElements(GLenum mode, GLsizei count, GLenum type, + const GLvoid *indices) +{ + GET_CURRENT_CONTEXT(ctx); + + if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices )) + return; - vbo_exec_DrawRangeElements(mode, min_index, max_index, count, type, indices); + vbo_validated_drawrangeelements(ctx, mode, GL_FALSE, ~0, ~0, + count, type, indices); } diff --git a/src/mesa/vbo/vbo_exec_draw.c b/src/mesa/vbo/vbo_exec_draw.c index 18419928b26..d76c45f3560 100644 --- a/src/mesa/vbo/vbo_exec_draw.c +++ b/src/mesa/vbo/vbo_exec_draw.c @@ -214,7 +214,7 @@ vbo_exec_bind_arrays( GLcontext *ctx ) /* override the default array set above */ exec->vtx.inputs[attr] = &arrays[attr]; - if (exec->vtx.bufferobj->Name) { + if (_mesa_is_bufferobj(exec->vtx.bufferobj)) { /* a real buffer obj: Ptr is an offset, not a pointer*/ GLsizeiptr offset; assert(exec->vtx.bufferobj->Pointer); /* buf should be mapped */ @@ -251,7 +251,7 @@ vbo_exec_vtx_unmap( struct vbo_exec_context *exec ) { GLenum target = GL_ARRAY_BUFFER_ARB; - if (exec->vtx.bufferobj->Name) { + if (_mesa_is_bufferobj(exec->vtx.bufferobj)) { GLcontext *ctx = exec->ctx; if (ctx->Driver.FlushMappedBufferRange) { @@ -291,7 +291,7 @@ vbo_exec_vtx_map( struct vbo_exec_context *exec ) MESA_MAP_NOWAIT_BIT; const GLenum usage = GL_STREAM_DRAW_ARB; - if (exec->vtx.bufferobj->Name == 0) + if (!_mesa_is_bufferobj(exec->vtx.bufferobj)) return; if (exec->vtx.buffer_map != NULL) { @@ -365,7 +365,7 @@ vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap ) if (ctx->NewState) _mesa_update_state( ctx ); - if (exec->vtx.bufferobj->Name) { + if (_mesa_is_bufferobj(exec->vtx.bufferobj)) { vbo_exec_vtx_unmap( exec ); } @@ -378,12 +378,13 @@ vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap ) exec->vtx.prim, exec->vtx.prim_count, NULL, + GL_TRUE, 0, exec->vtx.vert_count - 1); /* If using a real VBO, get new storage -- unless asked not to. */ - if (exec->vtx.bufferobj->Name && !unmap) { + if (_mesa_is_bufferobj(exec->vtx.bufferobj) && !unmap) { vbo_exec_vtx_map( exec ); } } @@ -392,7 +393,7 @@ vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap ) /* May have to unmap explicitly if we didn't draw: */ if (unmap && - exec->vtx.bufferobj->Name && + _mesa_is_bufferobj(exec->vtx.bufferobj) && exec->vtx.buffer_map) { vbo_exec_vtx_unmap( exec ); } diff --git a/src/mesa/vbo/vbo_rebase.c b/src/mesa/vbo/vbo_rebase.c index ea87dede646..3bf7ef580fc 100644 --- a/src/mesa/vbo/vbo_rebase.c +++ b/src/mesa/vbo/vbo_rebase.c @@ -208,6 +208,7 @@ void vbo_rebase_prims( GLcontext *ctx, prim, nr_prims, ib, + GL_TRUE, 0, max_index - min_index ); diff --git a/src/mesa/vbo/vbo_save_api.c b/src/mesa/vbo/vbo_save_api.c index d00d304d2ef..cdbbc9c1876 100644 --- a/src/mesa/vbo/vbo_save_api.c +++ b/src/mesa/vbo/vbo_save_api.c @@ -911,7 +911,7 @@ static void GLAPIENTRY _save_OBE_DrawElements(GLenum mode, GLsizei count, GLenum _ae_map_vbos( ctx ); - if (ctx->Array.ElementArrayBufferObj->Name) + if (_mesa_is_bufferobj(ctx->Array.ElementArrayBufferObj)) indices = ADD_POINTERS(ctx->Array.ElementArrayBufferObj->Pointer, indices); vbo_save_NotifyBegin( ctx, mode | VBO_SAVE_PRIM_WEAK ); diff --git a/src/mesa/vbo/vbo_save_draw.c b/src/mesa/vbo/vbo_save_draw.c index 5110648c282..d834fa1f2e7 100644 --- a/src/mesa/vbo/vbo_save_draw.c +++ b/src/mesa/vbo/vbo_save_draw.c @@ -279,6 +279,7 @@ void vbo_save_playback_vertex_list( GLcontext *ctx, void *data ) node->prim, node->prim_count, NULL, + GL_TRUE, 0, /* Node is a VBO, so this is ok */ node->count - 1); } diff --git a/src/mesa/vbo/vbo_split_copy.c b/src/mesa/vbo/vbo_split_copy.c index d7ffebf6076..8ec180d5508 100644 --- a/src/mesa/vbo/vbo_split_copy.c +++ b/src/mesa/vbo/vbo_split_copy.c @@ -30,6 +30,7 @@ */ #include "main/glheader.h" +#include "main/bufferobj.h" #include "main/imports.h" #include "main/image.h" #include "main/macros.h" @@ -194,6 +195,7 @@ flush( struct copy_context *copy ) copy->dstprim, copy->dstprim_nr, ©->dstib, + GL_TRUE, 0, copy->dstbuf_nr ); @@ -443,7 +445,7 @@ replay_init( struct copy_context *copy ) copy->varying[j].size = attr_size(copy->array[i]); copy->vertex_size += attr_size(copy->array[i]); - if (vbo->Name && !vbo->Pointer) + if (_mesa_is_bufferobj(vbo) && !_mesa_bufferobj_mapped(vbo)) ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER, GL_READ_ONLY, vbo); copy->varying[j].src_ptr = ADD_POINTERS(vbo->Pointer, @@ -457,7 +459,8 @@ replay_init( struct copy_context *copy ) * caller convert non-indexed prims to indexed. Could alternately * do it internally. */ - if (copy->ib->obj->Name && !copy->ib->obj->Pointer) + if (_mesa_is_bufferobj(copy->ib->obj) && + !_mesa_bufferobj_mapped(copy->ib->obj)) ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, GL_READ_ONLY, copy->ib->obj); @@ -561,13 +564,14 @@ replay_finish( struct copy_context *copy ) */ for (i = 0; i < copy->nr_varying; i++) { struct gl_buffer_object *vbo = copy->varying[i].array->BufferObj; - if (vbo->Name && vbo->Pointer) + if (_mesa_is_bufferobj(vbo) && _mesa_bufferobj_mapped(vbo)) ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, vbo); } /* Unmap index buffer: */ - if (copy->ib->obj->Name && copy->ib->obj->Pointer) { + if (_mesa_is_bufferobj(copy->ib->obj) && + _mesa_bufferobj_mapped(copy->ib->obj)) { ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, copy->ib->obj); } } diff --git a/src/mesa/vbo/vbo_split_inplace.c b/src/mesa/vbo/vbo_split_inplace.c index 266bc56c826..da84eaa6ead 100644 --- a/src/mesa/vbo/vbo_split_inplace.c +++ b/src/mesa/vbo/vbo_split_inplace.c @@ -59,11 +59,11 @@ struct split_context { static void flush_vertex( struct split_context *split ) { GLuint min_index, max_index; + GLuint i; if (!split->dstprim_nr) return; - GLuint i; min_index = split->dstprim[0].start; max_index = min_index + split->dstprim[0].count - 1; @@ -85,6 +85,7 @@ static void flush_vertex( struct split_context *split ) split->dstprim, split->dstprim_nr, NULL, + GL_TRUE, min_index, max_index); |